1	# =====================================================================
     2	# _rdfacpi.awk: replace CPI-encoded RDFa markup in string.
     3	#
     4	# Copyright (c) 2009 Carlo Strozzi
     5	#
     6	# This program is free software; you can redistribute it and/or modify
     7	# it under the terms of the GNU General Public License as published by
     8	# the Free Software Foundation; version 2 dated June, 1991.
     9	#
    10	# This program is distributed in the hope that it will be useful,
    11	# but WITHOUT ANY WARRANTY; without even the implied warranty of
    12	# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    13	# GNU General Public License for more details.
    14	# 
    15	# You should have received a copy of the GNU General Public License 
    16	# along with this program; if not, write to the Free Software
    17	# Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
    18	#
    19	# =====================================================================
    20	# string _rdfacpi(string s)
    21	#
    22	# A number of considerations are in order regarding this function:
    23	#
    24	# Only line-level CPIs are used, i.e. no block-level (::x:)...(:x::) .
    25	#
    26	# Each supported type is handled explicitly. Top level elements
    27	# ("organization", "person", etc.) will determine how lower level
    28	# elements will be handled. That is, an "address" element at top
    29	# level will be handled differently from an "address" element found
    30	# at top level.
    31	#
    32	# Currently Google recognizes also the following structures, which this
    33	# function tries to support:
    34	#
    35	# 1) "organization" data embedded in "review".
    36	#
    37	# 2) "person" data embedded in "review".
    38	#
    39	# 3) "product" data embedded in "review".
    40	#
    41	# 4) "address" data embedded in either "organization" or "person".
    42	#
    43	# Up to two levels of nesting are supported, such as "address" inside
    44	# an "organization" inside a "review". More can be supported in the
    45	# future simply by using a multi-valued integer for "subrel" and more
    46	# than two work arrays for _mrwresponse() (currently "m" and "m1", see
    47	# the code below).
    48	#
    49	# This function implements RDFa as proposed by Google, which is not
    50	# exactly what the actual standard mandates, especially concerning
    51	# the use of common and well-understood vocabularies. Please see W3C's
    52	# reference docs at http://www.w3.org/MarkUp/2009/rdfa-for-html-authors .
    53	#
    54	# This local implementation supports also "geo" data, embedded into
    55	# either "organization" or "person", because it is considered as an
    56	# alternate form of "postal" address here. Note that although it may
    57	# seem tempting to allow geo data also as a child element of "address",
    58	# this is probably not to be done because, as I said, geographical
    59	# coordinates should rather be considered an *alternate* form of
    60	# address, not a property of of an ordinary postal address as they can
    61	# exist indipendently of the latter. Note also that geo-tagging is NOT
    62	# supported on "review" entries, although they may be about places to
    63	# visit etc. This limitation can be overcome by resorting to inline 
    64	# geo-tagging through Microformat's <abbr> elements. See _mfmtcpi.awk
    65	# for more info, as well as further down inside this source code.
    66	#
    67	# Finally, this program is complicated, so do expect bugs.
    68	#
    69	# =====================================================================
    70	#
    71	# Sample input:
    72	#
    73	# Sample misplaced entry; since it belongs into the "address" context,
    74	# which is not open yet, it is left unparsed:
    75	#
    76	# (:v-region WA:)
    77	#
    78	# Open an "organization" context:
    79	#
    80	# (:v-organization ACME Example 1 Inc.^45.123;9.456:)
    81	# (:v-telephone +1 (0)234 123456:)
    82	#
    83	# In the previous example note the presence of "^45.123;9.456"
    84	# immediately following "Komala Vilas". The caret "^" stands for
    85	# "Abbreviation" and can be used to define an abbreviation for the
    86	# associated name. The exact type of abbreviation depends on the pattern
    87	# of caracters that follow the caret itself. Currently only embedded
    88	# geo-tagging is supported, but more abbreviation types may be added
    89	# in the future. Only one single abbreviation is taken into account,
    90	# i.e. if a pattern such as "Text^stuff1^stuff2" is specified, only
    91	# "Text^stuff1" will be considered. This seems reasonable since having
    92	# multiple abbreviations for the same piece of text seems useless.
    93	# Since RDFa has no provisions for abbreviations, I use the same
    94	# syntax produced by the Microformat scanner. See the source code of
    95	# the latter for more info on this.
    96	#
    97	# Sample extra entry that does not parse as RDFa
    98	# and is left unparsed:
    99	#
   100	# (:vvvv aaa bbb:)
   101	#
   102	# Other stuff belonging into the current "organization" context:
   103	#
   104	# (:v-url <a href="http://www1.example.com">http://www1.example.com</a>:)
   105	#
   106	# Postal Address; since we have an "organization" context already open,
   107	# this is treated as that organization's address:
   108	#
   109	# (:v-address Some Street 1, 123:)
   110	# (:v-locality Some Town 1:)
   111	#
   112	# Entering the same element twice (see previous line) is probably incorrect
   113	# but the parser currently accepts it:
   114	#
   115	# (:v-locality Some Other Town 2:)
   116	#
   117	# Sample mistyped data (trailing ":" after "region") that will
   118	# be left unparsed:
   119	#
   120	# (:v-region: NJ:)
   121	#
   122	# Other components of the current "address" context follow:
   123	#
   124	# (:v-zip 54321:)
   125	# (:v-country USA:)
   126	#
   127	# Organization after an address:
   128	#
   129	# (:v-organization ACME Example 2 Inc.:)
   130	# (:v-telephone +2 (0)234 123456:)
   131	#
   132	# Sample "product" entry; this is meaningless inside the current "address"
   133	# context", so it is sent one level above first, which is "organization",
   134	# and since it is invalid there too, a new top-level "product" context
   135	# is started:
   136	#
   137	# (:v-product Gizmo:)
   138	# (:v-cat Gadgets:)
   139	# (:v-desc A Nifty Useless Tingie:)
   140	# (:v-price 123.4:)
   141	# (:v-photo <a href="http://www.example.com/photos/p25.jpg">http://www.example.com/photos/p25.jpg</a>:)
   142	# (:v-url <a href="http://www2.example.com/products/p25">http://www2.example.com/products/p25</a>:)
   143	#
   144	# Sample "review" entry; this is meaningless inside the previous "product"
   145	# context, so a new top-level "review" context is started (reviews are
   146	# toplevel-only elements):
   147	#
   148	# (:v-review Komala Vilas 1:)
   149	# (:v-by Meenakshi Ammal:)
   150	# (:v-rating 1.1:)
   151	# (:v-date 1st April 2005:)
   152	# (:v-summary Best vegetarian food 1:)
   153	#
   154	# Sample "review" entry, with optional URI/URL/CURIE that will be used to
   155	# build the RDF "about" property of the entry (this is currently
   156	# supported by Google only for Reviews and Review-aggregates):
   157	#
   158	# (:v-review http://www.example.com Komala Vilas 2:)
   159	# (:v-by Meenakshi Ammal:)
   160	# (:v-rating 2.2:)
   161	# (:v-date 2nd April 2005:)
   162	# (:v-summary Best vegetarian food 2:)
   163	#
   164	# Sample "review" entry, with optional blank-node CURIE:
   165	#
   166	# (:v-review [_:KomalaResort] Komala Vilas 3:)
   167	# (:v-by Meenakshi Ammal:)
   168	# (:v-rating 3.3:)
   169	# (:v-date 3rd April 2005:)
   170	# (:v-summary Best vegetarian food 3:)
   171	#
   172	# Sample "review aggregate" (this one too supports optional URI/URL/CURIE):
   173	#
   174	# (:v-areview Komala Vilas 4:)
   175	# (:v-rating 3.7:)
   176	# (:v-count 20:)
   177	# (:v-summary Best vegetarian food 4:)
   178	#
   179	# Sample "review" entry, with the reviewer being an organization instead
   180	# of simple #CDATA with the name of the author; note also the geographical
   181	# coordinates:
   182	#
   183	# (:v-review Komala Vilas 5:)
   184	# (:v-organization ACME Example 3 Inc.:)
   185	# (:v-telephone +3 (0)234 123456:)
   186	# (:v-geo:)
   187	# (:v-latitude 45.123:)
   188	# (:v-longitude 9.456:)
   189	# (:v-address Some Street 2, 123:)
   190	# (:v-locality Some Town 3:)
   191	# (:v-rating 3.8:)
   192	# (:v-date 4th April 2005:)
   193	# (:v-summary Best vegetarian food 5:)
   194	#
   195	# Sample "review", again by an organization, where "v-telephone" has
   196	# been placed after the last element of the "address" context ("v-locality"):
   197	#
   198	# (:v-review Komala Vilas 6:)
   199	# (:v-organization ACME Example 4 Inc.:)
   200	# (:v-address Some Street 3, 123:)
   201	# (:v-locality Some Town 4:)
   202	# (:v-telephone +4 (0)234 123456:)
   203	# (:v-rating 3.9:)
   204	# (:v-date 5th April 2005:)
   205	# (:v-summary Best vegetarian food 6:)
   206	#
   207	# When the parser meets "v-telephone" in the "address" context" it
   208	# will bail-out from that context and move up one level, which is
   209	# the "organization" context. Since "v-telephone" is valid in that
   210	# context, things will parse normally.
   211	#
   212	# In the following example, "v-telephone" is misplaced after "v-rating":
   213	#
   214	# (:v-review Komala Vilas 7:)
   215	# (:v-organization ACME Example 5 Inc.:)
   216	# (:v-address Some Street 4, 123:)
   217	# (:v-locality Some Town 5:)
   218	# (:v-rating 4.0:)
   219	# (:v-telephone +5 (0)234 123456:)
   220	# (:v-date 6th April 2005:)
   221	# (:v-summary Best vegetarian food 7:)
   222	#
   223	# When the parser meets "v-rating" in the "address" context it bails-out
   224	# to the parent "organization" context, and since "v-rating" is not valid
   225	# in that context either, it bails-out further to the "review" context,
   226	# where "v-rating" is valid. Then, while in the "review" context, the parser
   227	# meets "v-telephone", which is invalid in that context, and the bail-out
   228	# process starts again, with the parser climbing backwards until a valid
   229	# element/context combination is met. Up to then, all elements encountered
   230	# along the way will be left unparsed. Since no elements in the example
   231	# belong into any of the upper contexts, they will all be left unparsed
   232	# up to the end of the "review" context.
   233	#
   234	# Entering two adjacent address blocks is probably incorrect,
   235	# but the parser currently accepts them:
   236	#
   237	# (:v-review Komala Vilas 8:)
   238	# (:v-organization ACME Example 6 Inc.:)
   239	# (:v-address Some Street 5, 123:)
   240	# (:v-locality Some Town 6:)
   241	# (:v-address Some Other Street 6, 123:)
   242	# (:v-locality Some Other Town 7:)
   243	#
   244	# In the next example a review done by an organization is followed by
   245	# a "person" block. Since reviews can take also persons as reviewers,
   246	# a "person" occurring while in a "review" context is considered a
   247	# reviewer. In this example therefore we will end up with having
   248	# two reviewers for the same review, that is an organization and a
   249	# person:
   250	#
   251	# (:v-review Komala Vilas 9^45.123;9.456:)
   252	# (:v-organization ACME Example 7 Inc.:)
   253	# (:v-telephone +6 (0)234 123456:)
   254	# (:v-address Some Street 7, 123:)
   255	# (:v-locality Some Town 8:)
   256	# (:v-rating 4.1:)
   257	# (:v-date 7th April 2005:)
   258	# (:v-summary Best vegetarian food 8:)
   259	#
   260	# (:v-person John Smith Jr. 1:)
   261	# (:v-role General Manager 1:)
   262	#
   263	# Having multiple reviewers in the same "review" context may be wrong for
   264	# Google's crawler, but the parser currently accepts it. If considering
   265	# "John Smith Jr." a second reviewer of "Komala Vilas" is not what we
   266	# mean, we must ensure to tell the parser by inserting a "v-break" element
   267	# between the end of the review and the beginning of the person:
   268	#
   269	# (:v-review Komala Vilas 10:)
   270	# (:v-organization ACME Example 8 Inc.:)
   271	# (:v-telephone +7 (0)234 123456:)
   272	# (:v-address Some Street 8, 123:)
   273	# (:v-locality Some Town 9:)
   274	# (:v-rating 4.2:)
   275	# (:v-date 8th April 2005:)
   276	# (:v-summary Best vegetarian food 9:)
   277	#
   278	# (:v-break:)
   279	#
   280	# (:v-person John Smith Jr. 2:)
   281	# (:v-role General Manager 2:)
   282	#
   283	#
   284	# Sample "person" entry, with an embedded address block:
   285	#
   286	# (:v-person John Smith Jr. 3:)
   287	# (:v-url <a href="http://www3.example.org/~jsmith">http://www.example.org/~jsmith</a>:)
   288	# (:v-photo <a href="http://www.example.org/~jsmith/photo.jpg">http://www.example.org/~jsmith/photo.jpg</a>:)
   289	# (:v-address Some Street 9, 892:)
   290	# (:v-locality Some Town 10:)
   291	# (:v-region NY:)
   292	# (:v-zip 56671:)
   293	# (:v-country USA:)
   294	#
   295	# Sample XFN properties of the current "person" context:
   296	# (:v-xfn friend <a href="http://brad-log1.example.org/">Brad 1</a>:)
   297	#
   298	# Same as above, but with multi-valued 'rel' attribute, which is supported
   299	# by the XFN specs but not by Google, so I omit the "v:" namespace prefix
   300	# on output:
   301	# (:v-xfn friend,met <a href="http://brad-log2.example.org/">Brad 2</a>:)
   302	#
   303	# Another XFN properties of the current "person", in a different format:
   304	# (:v-xfn acquaintance http://brad-log3.example.org Brad3:)
   305	#
   306	# Same as above, but with multi-valued 'rel' attribute, which is supported
   307	# by the XFN specs but not by Google, so I omit the "v:" namespace prefix
   308	# on output:
   309	# (:v-xfn co-worker,met http://brad-log4.example.org Brad4:)
   310	#
   311	# (:v-xfn me <a href="http://www.linux.it/~carlos/">Carlo</a>:)
   312	# 
   313	# In the above examples all tags have been written on a separate line
   314	# and without any interspersed text. That was done for readability, but
   315	# the whole purpose of the RDFa markup is the semantic labelling of
   316	# free-form text, so in practical situations a "person" definition
   317	# will be surrounded by- and interspersed with text, like this:
   318	#
   319	# Lorem ipsum dolor sit amet (:v-person John Smith Jr. 4:), consectetur
   320	# adipiscing elit (:v-role General Manager 3:). Aenean convallis
   321	# scelerisque metus, ...
   322	#
   323	# Following is the parsed RDFa output corresponding to the above input
   324	# with the interspersed running text removed for clarity:
   325	#
   326	# <div xmlns:v="http://rdf.data-vocabulary.org/#"
   327	# typeof="v:Organization">
   328	#   <span property="v:name">
   329	#     <abbr class="geo" title="45.123;9.456">ACME Example 1
   330	#     Inc.</abbr>
   331	#   </span>
   332	#   <span property="v:tel">+1 (0)234 123456</span>
   333	#   <span rel="v:url">
   334	#     <a href="http://www1.example.com">http://www1.example.com</a>
   335	#   </span>
   336	#   <span rel="v:address">
   337	#     <span property="v:street-address">Some Street 1, 123</span>
   338	#     <span property="v:locality">Some Town 1</span>
   339	#     <span property="v:locality">Some Other Town 2</span>
   340	#     <span property="v:postal-code">54321</span>
   341	#     <span property="v:country-name">USA</span>
   342	#   </span>
   343	# </div>
   344	# <div xmlns:v="http://rdf.data-vocabulary.org/#"
   345	# typeof="v:Organization">
   346	#   <span property="v:name">ACME Example 2 Inc.</span>
   347	#   <span property="v:tel">+2 (0)234 123456</span>
   348	# </div>
   349	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Product">
   350	#   <span property="v:name">Gizmo</span>
   351	#   <span property="v:category">Gadgets</span>
   352	#   <span property="v:description">A Nifty Useless Tingie</span>
   353	#   <span property="v:price">123.40</span>
   354	#   <span rel="v:photo">
   355	#     <a href="http://www.example.com/photos/p25.jpg">
   356	#     http://www.example.com/photos/p25.jpg</a>
   357	#   </span>
   358	#   <span rel="v:url">
   359	#     <a href="http://www2.example.com/products/p25">
   360	#     http://www2.example.com/products/p25</a>
   361	#   </span>
   362	# </div>
   363	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Review">
   364	#   <span property="v:itemReviewed">Komala Vilas 1</span>
   365	#   <span property="v:reviewer">Meenakshi Ammal</span>
   366	#   <span property="v:rating">1.1</span>
   367	#   <span property="v:dtReviewed">1st April 2005</span>
   368	#   <span property="v:summary">Best vegetarian food 1</span>
   369	# </div>
   370	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Review">
   371	#   <span rel="v:itemReviewed">
   372	#     <span about="http://www.example.com" property="v:name">Komala
   373	#     Vilas 2</span>
   374	#     <span property="v:reviewer">Meenakshi Ammal</span>
   375	#     <span property="v:rating">2.2</span>
   376	#     <span property="v:dtReviewed">2nd April 2005</span>
   377	#     <span property="v:summary">Best vegetarian food 2</span>
   378	#   </span>
   379	# </div>
   380	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Review">
   381	#   <span rel="v:itemReviewed">
   382	#     <span about="[_:KomalaResort]" property="v:name"
   383	#     rel="foaf:isPrimaryTopicOf">Komala Vilas 3</span>
   384	#     <span property="v:reviewer">Meenakshi Ammal</span>
   385	#     <span property="v:rating">3.3</span>
   386	#     <span property="v:dtReviewed">3rd April 2005</span>
   387	#     <span property="v:summary">Best vegetarian food 3</span>
   388	#   </span>
   389	# </div>
   390	# <div xmlns:v="http://rdf.data-vocabulary.org/#"
   391	# typeof="v:Review-aggregate">
   392	#   <span property="v:itemReviewed">Komala Vilas 4</span>
   393	#   <span rel="v:rating">
   394	#     <span property="v:average">3.7</span>
   395	#   </span>
   396	#   <span property="v:count">20</span>
   397	#   <span property="v:summary">Best vegetarian food 4</span>
   398	# </div>
   399	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Review">
   400	#   <span property="v:itemReviewed">Komala Vilas 5</span>
   401	#   <span rel="v:reviewer">
   402	#     <span typeof="v:Organization">
   403	#       <span property="v:name">ACME Example 3 Inc.</span>
   404	#       <span property="v:tel">+3 (0)234 123456</span>
   405	#       <span xmlns:geo="http://www.w3.org/2003/01/geo/wgs84_pos#"
   406	#       rel="geo:Point">
   407	#         <span property="geo:lat">45.123</span>
   408	#         <span property="geo:long">9.456</span>
   409	#       </span>
   410	#       <span rel="v:address">
   411	#         <span property="v:street-address">Some Street 2, 123</span>
   412	#         <span property="v:locality">Some Town 3</span>
   413	#       </span>
   414	#     </span>
   415	#   </span>
   416	#   <span property="v:rating">3.8</span>
   417	#   <span property="v:dtReviewed">4th April 2005</span>
   418	#   <span property="v:summary">Best vegetarian food 5</span>
   419	# </div>
   420	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Review">
   421	#   <span property="v:itemReviewed">Komala Vilas 6</span>
   422	#   <span rel="v:reviewer">
   423	#     <span typeof="v:Organization">
   424	#       <span property="v:name">ACME Example 4 Inc.</span>
   425	#       <span rel="v:address">
   426	#         <span property="v:street-address">Some Street 3, 123</span>
   427	#         <span property="v:locality">Some Town 4</span>
   428	#       </span>
   429	#       <span property="v:tel">+4 (0)234 123456</span>
   430	#     </span>
   431	#   </span>
   432	#   <span property="v:rating">3.9</span>
   433	#   <span property="v:dtReviewed">5th April 2005</span>
   434	#   <span property="v:summary">Best vegetarian food 6</span>
   435	# </div>
   436	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Review">
   437	# <span property="v:itemReviewed">Komala Vilas 7</span>
   438	# <span rel="v:reviewer">
   439	#   <span typeof="v:Organization">
   440	#     <span property="v:name">ACME Example 5 Inc.</span>
   441	#     <span rel="v:address">
   442	#       <span property="v:street-address">Some Street 4, 123</span>
   443	#       <span property="v:locality">Some Town 5</span>
   444	#     </span>
   445	#   </span>
   446	# </span>
   447	# <span property="v:rating">4.0</span>(:v-telephone +5 (0)234
   448	# 123456:)(:v-date 6th April 2005:)(:v-summary Best vegetarian food
   449	# 7:)</div>
   450	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Review">
   451	#   <span property="v:itemReviewed">Komala Vilas 8</span>
   452	#   <span rel="v:reviewer">
   453	#     <span typeof="v:Organization">
   454	#       <span property="v:name">ACME Example 6 Inc.</span>
   455	#       <span rel="v:address">
   456	#         <span property="v:street-address">Some Street 5, 123</span>
   457	#         <span property="v:locality">Some Town 6</span>
   458	#       </span>
   459	#       <span rel="v:address">
   460	#         <span property="v:street-address">Some Other Street 6,
   461	#         123</span>
   462	#         <span property="v:locality">Some Other Town 7</span>
   463	#       </span>
   464	#     </span>
   465	#   </span>
   466	# </div>
   467	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Review">
   468	#   <span property="v:itemReviewed">
   469	#     <abbr class="geo" title="45.123;9.456">Komala Vilas 9</abbr>
   470	#   </span>
   471	#   <span rel="v:reviewer">
   472	#     <span typeof="v:Organization">
   473	#       <span property="v:name">ACME Example 7 Inc.</span>
   474	#       <span property="v:tel">+6 (0)234 123456</span>
   475	#       <span rel="v:address">
   476	#         <span property="v:street-address">Some Street 7, 123</span>
   477	#         <span property="v:locality">Some Town 8</span>
   478	#       </span>
   479	#     </span>
   480	#   </span>
   481	#   <span property="v:rating">4.1</span>
   482	#   <span property="v:dtReviewed">7th April 2005</span>
   483	#   <span property="v:summary">Best vegetarian food 8</span>
   484	#   <span rel="v:reviewer">
   485	#     <span typeof="v:Person">
   486	#       <span property="v:name">John Smith Jr. 1</span>
   487	#       <span property="v:role">General Manager 1</span>
   488	#     </span>
   489	#   </span>
   490	# </div>
   491	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Review">
   492	#   <span property="v:itemReviewed">Komala Vilas 10</span>
   493	#   <span rel="v:reviewer">
   494	#     <span typeof="v:Organization">
   495	#       <span property="v:name">ACME Example 8 Inc.</span>
   496	#       <span property="v:tel">+7 (0)234 123456</span>
   497	#       <span rel="v:address">
   498	#         <span property="v:street-address">Some Street 8, 123</span>
   499	#         <span property="v:locality">Some Town 9</span>
   500	#       </span>
   501	#     </span>
   502	#   </span>
   503	#   <span property="v:rating">4.2</span>
   504	#   <span property="v:dtReviewed">8th April 2005</span>
   505	#   <span property="v:summary">Best vegetarian food 9</span>
   506	# </div>
   507	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Person">
   508	#   <span property="v:name">John Smith Jr. 2</span>
   509	#   <span property="v:role">General Manager 2</span>
   510	# </div>
   511	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Person">
   512	#   <span property="v:name">John Smith Jr. 3</span>
   513	#   <span rel="v:url">
   514	#     <a href="http://www3.example.org/~jsmith">
   515	#     http://www.example.org/~jsmith</a>
   516	#   </span>
   517	#   <span rel="v:photo">
   518	#     <a href="http://www.example.org/~jsmith/photo.jpg">
   519	#     http://www.example.org/~jsmith/photo.jpg</a>
   520	#   </span>
   521	#   <span rel="v:address">
   522	#     <span property="v:street-address">Some Street 9, 892</span>
   523	#     <span property="v:locality">Some Town 10</span>
   524	#     <span property="v:region">NY</span>
   525	#     <span property="v:postal-code">56671</span>
   526	#     <span property="v:country-name">USA</span>
   527	#   </span>
   528	#   <a rel='friend' href="http://brad-log1.example.org/">Brad 1</a>
   529	#   <a rel='friend met' href="http://brad-log2.example.org/">Brad
   530	#   2</a>
   531	#   <a href="http://brad-log3.example.org" rel="acquaintance">
   532	#   Brad3</a>
   533	#   <a href="http://brad-log4.example.org" rel="co-worker met">
   534	#   Brad4</a>
   535	#   <a rel='me' href="http://www.linux.it/~carlos/">Carlo</a>
   536	# </div>
   537	# <div xmlns:v="http://rdf.data-vocabulary.org/#" typeof="v:Person">
   538	#   <span property="v:name">John Smith Jr. 4</span>
   539	#   <span property="v:role">General Manager 3</span>
   540	# </div>
   541	#
   542	# =====================================================================
   543	#
   544	# See also the following additional sources:
   545	#
   546	# * The source code of the _mrwresponse() CSA function.
   547	# * http://en.wikipedia.org/wiki/RDFa
   548	# * http://linuxgazette.net/105/misc/oregan/foaf-example.rdf.txt
   549	# * http://www.google.com/support/webmasters/bin/answer.py?answer=146898
   550	# * http://en.wikipedia.org/wiki/CURIE
   551	# * http://www.w3.org/MarkUp/2009/rdfa-for-html-authors (important!)
   552	# * http://gmpg.org/xfn/intro
   553	# * http://gmpg.org/xfn/background (this is *key* to understanding XFN)
   554	# * http://www.gmpg.org/xfn/join
   555	# * http://rubhub.com/main/add
   556	# * http://www.w3.org/2003/01/geo/
   557	#
   558	# This link shows how to easily get long/lat from a Google Map search:
   559	# http://lifehacker.com/267361/how-to-find-latitude-and-longitude
   560	#
   561	# The W3C validation service at http://validator.w3.org/check can be
   562	# used to validtate RDFa documents, but it requires the document itself
   563	# to be strictly conformant with DOCTYPE XHTML+RDFa 1.0 , i.e. it is
   564	# expected to begin with:
   565	#
   566	# <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML+RDFa 1.0//EN"
   567	#	"http://www.w3.org/MarkUp/DTD/xhtml-rdfa-1.dtd">
   568	#
   569	# This strict conformancy may be problematic to attain, and it can be
   570	# overkill if all we need is to test our embedded RDFa statements.
   571	# A more "quick-and-dirty" way to see if our RDFa code is OK is to use
   572	# W3C's RDFa Distiller Service at http://www.w3.org/2007/08/pyRdfa/ .
   573	#
   574	# Where appropriate, this function supports "about" properties in the
   575	# form of URLs, URIs, CURIEs and blank-node (bnode) CURIEs.
   576	# =====================================================================
   577	
   578	function _rdfacpi(s,	   a,b,c,scantype,rel,name,value,uri,tag,S,\
   579				   re,oldtype,i,j,saved,flag,m,m1,subrel,div) {
   580	
   581	  # Get local settings and apply defaults.
   582	  if ((div=_rcget("TNS_GROUP_MISC_PROP",3)) != "span") div = "div"
   583	
   584	  delete a; delete b; delete c; delete m; delete m1
   585	
   586	  # Pupulate array of valid tags. Note the presence of the "geo"
   587	  # tag, currently not part of Google's specs and accepted only for
   588	  # "organization" and "review", as organizations usually stand at a
   589	  # fixed geographical location, and reviews may be about a place to
   590	  # visit, etc. See also http://www.w3.org/2003/01/geo/ .
   591	
   592	  tag["addr"] = 1
   593	  tag["address"] = 1
   594	  tag["affiliation"] = 1
   595	  tag["areview"] = 1
   596	  tag["br"] = 1
   597	  tag["brand"] = 1
   598	  tag["break"] = 1
   599	  tag["by"] = 1
   600	  tag["cat"] = 1
   601	  tag["category"] = 1
   602	  tag["count"] = 1
   603	  tag["country"] = 1
   604	  tag["date"] = 1
   605	  tag["desc"] = 1
   606	  tag["description"] = 1
   607	  tag["geo"] = 1
   608	  tag["lat"] = 1
   609	  tag["latitude"] = 1
   610	  tag["loc"] = 1
   611	  tag["locality"] = 1
   612	  tag["long"] = 1
   613	  tag["longitude"] = 1
   614	  tag["nick"] = 1
   615	  tag["nickname"] = 1
   616	  tag["org"] = 1
   617	  tag["organization"] = 1
   618	  tag["person"] = 1
   619	  tag["photo"] = 1
   620	  tag["price"] = 1
   621	  tag["prod"] = 1
   622	  tag["product"] = 1
   623	  tag["rating"] = 1
   624	  tag["region"] = 1
   625	  tag["review"] = 1
   626	  tag["reviewer"] = 1
   627	  tag["role"] = 1
   628	  tag["summary"] = 1
   629	  tag["tel"] = 1
   630	  tag["telephone"] = 1
   631	  tag["title"] = 1
   632	  tag["url"] = 1
   633	  tag["xfn"] = 1
   634	  tag["zip"] = 1
   635	
   636	  gsub(/:\)/,"\001",s)
   637	
   638	  while (_extract(s,"\\(:v-[^\\001]+\\001")) {
   639	
   640	      # Better comment this out while debugging.
   641	      S = S _RLMATCH
   642	
   643	      saved = _RMATCH
   644	
   645	      sub(/^\(:v-/,"",_RMATCH)
   646	      sub(/\001/,"",_RMATCH)
   647	      name = value = _RMATCH
   648	      name = tolower(name)
   649	      sub(/ .*/,"",name)
   650	
   651	      if (!tag[name]) {
   652		 # Better comment this out while debugging.
   653		 S = S "(:v-" _RMATCH "\001"
   654		 continue
   655	      }
   656	
   657	      if (!sub(/^[^ ]+ +/,"",value)) value = _NULL
   658	
   659	      # Handle each supported type explicitly.
   660	
   661	      # review and review-aggregate.
   662	      if (name ~ /^(a?review|br(eak)?)$/) {
   663	
   664		 # Try and detect optional URI in entity name. This is currently
   665		 # of (possibly future) use only with "review" types and person's
   666		 # affiliation, as per the relevant Google docs.
   667	
   668		 uri = value
   669		 if (sub(/ .*/,"",uri) && _isuri(uri) == _TRUE)
   670						sub(/^[^ ]+ +/,"",value)
   671		 else uri = ""
   672	
   673		 # Like any toplevel-only entries, [a]reviews always bring us
   674		 # back to top level. If by mistake they occur at a lower level
   675		 # the remaining entries of such lower level(s) will either be
   676		 # printed verbatim or they will become part of other aggregates
   677		 # if compatible. While reviews accept nested "Person" or
   678		 # "Organization" as the reviewer, review-aggregates do not,
   679		 # as it would not make sense because each review is (or should
   680		 # be) from a different source.
   681	
   682		 if (subrel) S = S _mrwresponse(m1,".","","","","","","",a,b,1,"!")
   683	
   684		 S = S _mrwresponse(m,".","","","","","","",a,b,1,"!")
   685	
   686		 scantype = name
   687		 delete oldtype
   688		 i = j = rel = 0
   689	
   690		 # Handle explicit break requests.
   691		 if (name ~ /^br/) {
   692		    scantype = ""
   693		    continue
   694		 }
   695	
   696		 oldtype[++i] = scantype
   697	      }
   698	
   699	      # organization, possibly within review.
   700	      else if (name ~ /^org(anization)?$/ && \
   701		(scantype == "" || scantype == "review")) {
   702		scantype == "" ? rel = 0 : rel = 1
   703		oldtype[++i] = scantype
   704		scantype = "org"
   705	      }
   706	
   707	      # product, possibly within review.
   708	      else if (name ~ /^prod(uct)?$/ && \
   709		(scantype == "" || scantype == "review")) {
   710		scantype == "" ? rel = 0 : rel = 1
   711		oldtype[++i] = scantype
   712		scantype = "prod"
   713	      }
   714	
   715	      # person, possibly within review.
   716	      else if (name == "person" && \
   717		(scantype == "" || scantype == "review")) {
   718		scantype == "" ? rel = 0 : rel = 1
   719		oldtype[++i] = scantype
   720		scantype = "person"
   721	      }
   722	
   723	      # address (i.e. place) , possibly within either "person" or
   724	      # "organization".
   725	      else if (name ~ /^addr(ess)?$/ && \
   726		(scantype == "" || scantype == "person" || scantype == "org")) {
   727		scantype == "" ? rel = 0 : rel = 1
   728		oldtype[++i] = scantype
   729		scantype = "addr"
   730	      }
   731	
   732	      # geographical coordinates, possibly within either "organization"
   733	      # or "person".
   734	      else if (name == "geo" && \
   735		(scantype == "" || scantype == "org" || scantype == "person")) {
   736		scantype == "" ? rel = 0 : rel = 1
   737		oldtype[++i] = scantype
   738		scantype = "geo"
   739	      }
   740	
   741	      else if (flag) {
   742		if (subrel) S = S _mrwresponse(m1,"",".","","","","","",a,b,1,"!")
   743		else S = S _mrwresponse(m,"",".","","","","","",a,b,1,"!")
   744	        flag = subrel = 0
   745	      }
   746	
   747	      if (scantype == "org") {
   748	
   749		 # This must be tested on each loop!
   750		 if (oldtype[i] == "") rel = 0
   751	
   752		 flag = 0
   753	
   754	         if (name ~ /^org(anization)?$/) {
   755		    if (!rel) {
   756	     	       a[1] = "xmlns:v"
   757	     	       b[1] = "http://rdf.data-vocabulary.org/#"
   758	     	       a[2] = "typeof"
   759	     	       b[2] = "v:Organization"
   760		       S = S _mrwresponse(m,div,"","","","","","",a,b,1,"!")
   761		    }
   762		    else {
   763		       a[1] = "rel"
   764		       b[1] = "v:reviewer"	# organization within review
   765	     	       S = S _mrwresponse(m,"","span","","","","","",a,b,1,"!")
   766	
   767		       # The typeof attribute isn't optional here, as the
   768		       # reviewer can be either a Person or an Organization,
   769		       # so it must be stated explicitly to avoid ambiguity.
   770	
   771		       a[1] = "typeof"
   772		       b[1] = "v:Organization"
   773	     	       S = S _mrwresponse(m,"","","","","span","","",a,b,1,"!")
   774		    }
   775	
   776		    # Handle optional abbreviation for "v:name".
   777		    # The caret "^" resembles an "A" and stands for "Abbreviation".
   778		    if (split(value,c,"^") > 1) {
   779		       if (c[2] ~ /^[-+]?[0-9]+\.[0-9]+ *; *[-+]?[0-9]+\.[0-9]+/)
   780				value = "<abbr class=\"geo\" title=\"" \
   781							c[2] "\">" c[1] "</abbr>"
   782		    }
   783		    # Add more optional abbreviations below.
   784	
   785	     	    a[1] = "property"
   786	     	    b[1] = "v:name"
   787	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   788		 }
   789	         else if (name ~ /^tel(ephone)?$/) {
   790	     	    a[1] = "property"
   791	     	    b[1] = "v:tel"
   792	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   793	         }
   794	         else if (name == "url") {
   795	     	    a[1] = "rel"
   796	     	    b[1] = "v:url"
   797	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   798	         }
   799		 else {
   800		    scantype = oldtype[i--]
   801		    s = saved s				# no match, stack it back.
   802		    flag = 1
   803		 }
   804	      }
   805	
   806	      else if (scantype == "person") {
   807	
   808		 # This must be checked on each loop!
   809		 if (oldtype[i] == "") rel = 0
   810	
   811	         if (name == "person") {
   812		    if (!rel) {
   813	     	       a[1] = "xmlns:v"
   814	     	       b[1] = "http://rdf.data-vocabulary.org/#"
   815	     	       a[2] = "typeof"
   816	     	       b[2] = "v:Person"
   817		       S = S _mrwresponse(m,div,"","","","","","",a,b,1,"!")
   818		    }
   819		    else {
   820		       a[1] = "rel"
   821		       b[1] = "v:reviewer"	# person within review
   822	     	       S = S _mrwresponse(m,"","span","","","","","",a,b,1,"!")
   823	
   824		       # The typeof attribute isn't optional here, as the
   825		       # reviewer can be either a Person or an Organization,
   826		       # so it must be stated explicitly to avoid ambiguity.
   827	
   828		       a[1] = "typeof"
   829		       b[1] = "v:Person"
   830	     	       S = S _mrwresponse(m,"","","","","span","","",a,b,1,"!")
   831		    }
   832	
   833	     	    a[1] = "property"
   834	     	    b[1] = "v:name"
   835	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   836		 }
   837	         else if (name ~ /^nick(name)?$/) {
   838	     	    a[1] = "property"
   839	     	    b[1] = "v:nickname"
   840	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   841	         }
   842	         else if (name ~ /^(org|affiliation)$/){
   843	
   844		    # Try and detect optional URI in entry. This is currently of
   845		    # (possibly future) use only with "review" types and person's
   846		    # affiliation, as per the relevant Google docs.
   847	
   848		    uri = value
   849		    if (sub(/ .*/,"",uri) && _isuri(uri) == _TRUE)
   850						sub(/^[^ ]+ +/,"",value)
   851		    else uri = ""
   852	
   853		    if (uri == "") {
   854		       a[1] = "property"
   855		       b[1] = "v:affiliation"
   856		       S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   857		    }
   858		    else {
   859		       a[1] = "rel"
   860	     	       b[1] = "v:affiliation"
   861	     	       S = S _mrwresponse(m,"","span","","","","","",a,b,1,"!")
   862	
   863	     	       a[1] = "about"
   864	     	       b[1] = uri
   865		       a[2] = "property"
   866	     	       b[2] = "v:name"
   867	
   868		       # Detect also bnode CURIEs, as explained here:
   869		       # http://www.w3.org/MarkUp/2009/rdfa-for-html-authors
   870		       if (uri ~ /^\[_:/) {
   871		       	  a[4] = "rel"
   872		       	  b[4] = "foaf:isPrimaryTopicOf"
   873		       }
   874	     	       S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   875		    }
   876	         }
   877	         else if (name ~ /^(url|photo)$/) {
   878	     	    a[1] = "rel"
   879	     	    b[1] = "v:" name
   880	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   881	         }
   882	         else if (name ~ /^(title|role)$/) {
   883	     	    a[1] = "property"
   884	     	    b[1] = "v:" name
   885	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   886	         }
   887	
   888		 # Google also recognizes XFN's "friend", "contact" and
   889		 # "acquaintance" properties, which are used to identify
   890		 # social relationships in a grassroots fashion typical
   891		 # of the so-called "microformats", with XFN being a
   892		 # lightweight alternative to FOAF. For more info see
   893		 # http://gmpg.org/xfn/intro . This code tries and support
   894		 # a larger XFN subset than the one supported by Google.
   895		 # I make no validity checks on the type of XFN rel(s)
   896		 # specified by the user, so I can potentially support
   897		 # any of them, even future ones.
   898	
   899	         else if (name == "xfn") {
   900		    name = value = _strip(value,_O_CRUSH)
   901		    sub(/ .*/,"",name)
   902		    sub(/[^ ]+ /,"",value)
   903		    sub(/,/," ",name)
   904	
   905		    # Prepend the "v:" namespace prefix only to those
   906		    # (single) values supported by Google.
   907		    #if (name ~ /^(friend|contact|acquaintance)$/) name = "v:" name
   908	
   909		    if (value ~ /^<[aA] /) {
   910		       # append to existing rel='' if any, or set one anew.
   911		       re = "[^a-z]*rel=[\"']"
   912		       if (value !~ re "[^'\"]*" name && \
   913				!sub(re, "&" name " ",value))
   914				sub(/ /," rel='" name "' ",value)
   915		       S = S value
   916		    }
   917		    else if (split(value,b," ") == 2) {
   918		       b[3] = b[2]
   919		       a[1] = "href"
   920	     	       a[2] = "rel"
   921	     	       b[2] = name
   922	     	       S = S _mrwresponse(m,"","","","","","a",b[3],a,b,1,"!")
   923		    }
   924		    else {
   925		       scantype = oldtype[i--]
   926		       s = saved s			# invalid, stack it back.
   927		       flag = 1
   928		    }
   929	         }
   930		 else {
   931		    scantype = oldtype[i--]
   932		    s = saved s				# no match, stack it back.
   933		    flag = 1
   934		 }
   935	      }
   936	
   937	      else if (scantype == "addr") {
   938	
   939		 # This must be checked on each loop!
   940		 if (oldtype[i] == "") rel = 0
   941	
   942		 flag = 0
   943	
   944	         if (name ~ /^addr(ess)?$/) {
   945	
   946		    if (oldtype[i-1] ~ /^(org|person|review)$/) subrel = 1
   947		    else subrel = 0
   948	
   949		    if (!rel) {
   950	     	       a[1] = "xmlns:v"
   951	     	       b[1] = "http://rdf.data-vocabulary.org/#"
   952	     	       a[2] = "typeof"
   953	     	       b[2] = "v:PostalAddress"
   954		       S = S _mrwresponse(m,div,"","","","","","",a,b,1,"!")
   955		    }
   956		    else {
   957		       a[1] = "rel"
   958		       b[1] = "v:address"
   959	     	       if (subrel)
   960			  S = S _mrwresponse(m1,"","span","","","","","",a,b,1,"!")
   961		       else
   962			  S = S _mrwresponse(m,"","span","","","","","",a,b,1,"!")
   963	
   964		       # A "v:address" rel entry can only be an address type,
   965		       # so no explicity typeof property is necessary. I therefore
   966		       # omit it for brevity, see Google's RDFa specs.
   967		       #a[1] = "typeof"
   968		       #b[1] = "v:PostalAddress"
   969		       #if (subrel) ...
   970	     	       #else ...
   971		    }
   972	     	    a[1] = "property"
   973	     	    b[1] = "v:street-address"
   974	     	    if (subrel)
   975			S = S _mrwresponse(m1,"","","","","","span",value,a,b,1,"!",1)
   976	     	    else
   977			S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   978		 }
   979	         else if (name ~ /^loc(ality)?$/) {
   980	     	    a[1] = "property"
   981	     	    b[1] = "v:locality"
   982		    if (subrel)
   983			S = S _mrwresponse(m1,"","","","","","span",value,a,b,1,"!",1)
   984		    else
   985			S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   986	         }
   987	         else if (name == "region") {
   988	     	    a[1] = "property"
   989	     	    b[1] = "v:region"
   990		    if (subrel)
   991			S = S _mrwresponse(m1,"","","","","","span",value,a,b,1,"!",1)
   992		    else
   993			S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
   994	         }
   995	         else if (name == "country") {
   996	     	    a[1] = "property"
   997	     	    b[1] = "v:country-name"
   998		    if (subrel)
   999			S = S _mrwresponse(m1,"","","","","","span",value,a,b,1,"!",1)
  1000		    else
  1001			S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1002	         }
  1003	         else if (name == "zip") {
  1004	     	    a[1] = "property"
  1005	     	    b[1] = "v:postal-code"
  1006		    if (subrel)
  1007			S = S _mrwresponse(m1,"","","","","","span",value,a,b,1,"!",1)
  1008		    else
  1009			S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1010	         }
  1011		 else {
  1012		    scantype = oldtype[i--]
  1013		    s = saved s				# no match, stack it back.
  1014		    flag = 1
  1015		 }
  1016	      }
  1017	
  1018	      else if (scantype == "prod") {
  1019	
  1020		 # This must be checked on each loop!
  1021		 if (oldtype[i] == "") rel = 0
  1022	
  1023	         if (name ~ /^prod(uct)?$/) {
  1024		    if (!rel) {
  1025	     	       a[1] = "xmlns:v"
  1026	     	       b[1] = "http://rdf.data-vocabulary.org/#"
  1027	     	       a[2] = "typeof"
  1028	     	       b[2] = "v:Product"
  1029		       S = S _mrwresponse(m,div,"","","","","","",a,b,1,"!")
  1030		    }
  1031		    else {
  1032		       a[1] = "rel"
  1033		       b[1] = "v:itemReviewed"		# product within review
  1034	     	       S = S _mrwresponse(m,"","span","","","","","",a,b,1,"!")
  1035	
  1036		       a[1] = "typeof"
  1037		       b[1] = "v:Product"
  1038	     	       S = S _mrwresponse(m,"","","","","span","","",a,b,1,"!")
  1039		    }
  1040	     	    a[1] = "property"
  1041	     	    b[1] = "v:name"
  1042	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1043		 }
  1044	         else if (name == "brand") {
  1045	     	    a[1] = "property"
  1046	     	    b[1] = "v:brand"
  1047	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1048	         }
  1049	         else if (name ~ /^cat(egory)?$/) {
  1050	     	    a[1] = "property"
  1051	     	    b[1] = "v:category"
  1052	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1053	         }
  1054	         else if (name ~ /^desc(ription)?$/) {
  1055	     	    a[1] = "property"
  1056	     	    b[1] = "v:description"
  1057	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1058	         }
  1059	         else if (name == "price") {
  1060	     	    a[1] = "property"
  1061	     	    b[1] = "v:price"
  1062		    value = sprintf("%.2f",value/1)	# RDFa specs.
  1063	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1064	         }
  1065	         else if (name == "photo") {
  1066	     	    a[1] = "rel"
  1067	     	    b[1] = "v:photo"
  1068	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1069	         }
  1070	         else if (name == "url") {
  1071	     	    a[1] = "rel"
  1072	     	    b[1] = "v:url"
  1073	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1074	         }
  1075		 else {
  1076		    scantype = oldtype[i--]
  1077		    s = saved s				# no match, stack it back.
  1078		    flag = 1
  1079		 }
  1080	      }
  1081	
  1082	      else if (scantype == "review") {
  1083	
  1084	         if (name == "review") {
  1085	     	    a[1] = "xmlns:v"
  1086	     	    b[1] = "http://rdf.data-vocabulary.org/#"
  1087	     	    a[2] = "typeof"
  1088	     	    b[2] = "v:Review"
  1089		    S = S _mrwresponse(m,div,"","","","","","",a,b,1,"!")
  1090	
  1091		    # Handle optional abbreviation for "v:name".
  1092		    # The caret "^" resembles an "A" and stands for "Abbreviation".
  1093		    if (split(value,c,"^") > 1) {
  1094		       if (c[2] ~ /^[-+]?[0-9]+\.[0-9]+ *; *[-+]?[0-9]+\.[0-9]+/)
  1095				value = "<abbr class=\"geo\" title=\"" \
  1096							c[2] "\">" c[1] "</abbr>"
  1097		    }
  1098		    # Add more optional abbreviations below.
  1099	
  1100		    if (uri == "") {
  1101		       a[1] = "property"
  1102	     	       b[1] = "v:itemReviewed"
  1103	     	       S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1104		    }
  1105		    else {
  1106		       a[1] = "rel"
  1107	     	       b[1] = "v:itemReviewed"
  1108	     	       S = S _mrwresponse(m,"","span","","","","","",a,b,1,"!")
  1109	
  1110	     	       a[1] = "about"
  1111	     	       b[1] = uri
  1112		       a[2] = "property"
  1113	     	       b[2] = "v:name"
  1114	
  1115		       # Detect also bnode CURIEs, as explained here:
  1116		       # http://www.w3.org/MarkUp/2009/rdfa-for-html-authors
  1117		       if (uri ~ /^\[_:/) {
  1118		       	  a[4] = "rel"
  1119		       	  b[4] = "foaf:isPrimaryTopicOf"
  1120		       }
  1121	
  1122	     	       S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1123		    }
  1124		 }
  1125	         else if (name == "by" || name == "reviewer") {
  1126	     	    a[1] = "property"
  1127	     	    b[1] = "v:reviewer"
  1128	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1129	         }
  1130	         else if (name == "rating") {
  1131	     	    a[1] = "property"
  1132	     	    b[1] = "v:rating"
  1133	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1134	         }
  1135	         else if (name == "date") {
  1136	     	    a[1] = "property"
  1137	     	    b[1] = "v:dtReviewed"
  1138	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1139	         }
  1140	         else if (name == "summary") {
  1141	     	    a[1] = "property"
  1142	     	    b[1] = "v:summary"
  1143	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1144	         }
  1145	         else if (name ~ /^desc(ription)?$/) {
  1146	     	    a[1] = "property"
  1147	     	    b[1] = "v:description"
  1148	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1149	         }
  1150		 else {
  1151		    scantype = oldtype[i--]
  1152		    s = saved s				# no match, stack it back.
  1153		    flag = 1
  1154		 }
  1155	      }
  1156	
  1157	      else if (scantype == "areview") {
  1158	
  1159	         if (name == "areview") {
  1160	     	    a[1] = "xmlns:v"
  1161	     	    b[1] = "http://rdf.data-vocabulary.org/#"
  1162	     	    a[2] = "typeof"
  1163	     	    b[2] = "v:Review-aggregate"
  1164		    S = S _mrwresponse(m,div,"","","","","","",a,b,1,"!")
  1165	
  1166		    if (uri == "") {
  1167		       a[1] = "property"
  1168	     	       b[1] = "v:itemReviewed"
  1169	     	       S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1170		    }
  1171		    else {
  1172		       a[1] = "rel"
  1173	     	       b[1] = "v:itemReviewed"
  1174	     	       S = S _mrwresponse(m,"","span","","","","","",a,b,1,"!")
  1175	
  1176	     	       a[1] = "about"
  1177	     	       b[1] = uri
  1178		       a[2] = "property"
  1179	     	       b[2] = "v:name"
  1180	
  1181		       # Detect also bnode CURIEs, as explained here:
  1182		       # http://www.w3.org/MarkUp/2009/rdfa-for-html-authors
  1183		       if (uri ~ /^\[_:/) {
  1184		       	  a[4] = "rel"
  1185		       	  b[4] = "foaf:isPrimaryTopicOf"
  1186		       }
  1187	
  1188		       # Handle optional abbreviation for "v:name".
  1189		       # The caret "^" resembles an "A" and stands for "Abbreviation".
  1190		       if (split(value,c,"^") > 1) {
  1191			  if (c[2] ~ /^[-+]?[0-9]+\.[0-9]+ *; *[-+]?[0-9]+\.[0-9]+/)
  1192				value = "<abbr class=\"geo\" title=\"" \
  1193							c[2] "\">" c[1] "</abbr>"
  1194		       }
  1195		       # Add more optional abbreviations below.
  1196	
  1197	     	       S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1198		    }
  1199		 }
  1200	         else if (name == "rating") {
  1201	     	    a[1] = "rel"
  1202	     	    b[1] = "v:rating"
  1203	     	    S = S _mrwresponse(m,"","span","","","","",value,a,b,1,"!")
  1204	
  1205	     	    a[1] = "property"
  1206	     	    b[1] = "v:average"
  1207	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!")
  1208	     	    S = S _mrwresponse(m,"",".","","","","","",a,b,1,"!")
  1209	         }
  1210	         else if (name == "count") {
  1211	     	    a[1] = "property"
  1212	     	    b[1] = "v:count"
  1213	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1214	         }
  1215	         else if (name == "summary") {
  1216	     	    a[1] = "property"
  1217	     	    b[1] = "v:summary"
  1218	     	    S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1219	         }
  1220		 else {
  1221		    scantype = oldtype[i--]
  1222		    s = saved s				# no match, stack it back.
  1223		    flag = 1
  1224		 }
  1225	      }
  1226	
  1227	      else if (scantype == "geo") {
  1228	
  1229		 # This must be tested on each loop!
  1230		 if (oldtype[i] == "") rel = 0
  1231	
  1232		 flag = 0
  1233	
  1234	         if (name == "geo") {
  1235	
  1236		    if (oldtype[i-1] ~ /^(org|person|review)$/) subrel = 1
  1237		    else subrel = 0
  1238	
  1239		    if (!rel) {
  1240	     	       a[1] = "xmlns:v"
  1241		       b[1] = "http://rdf.data-vocabulary.org/#"
  1242	
  1243		       # Geo data is not supported by Google RDFa specs yet
  1244		       # so I need to point at a different namespace here.
  1245		       # See http://www.w3.org/2003/01/geo/#example .
  1246	
  1247		       a[2] = "xmlns:geo"
  1248		       b[2] = "http://www.w3.org/2003/01/geo/wgs84_pos#"
  1249	
  1250		       a[3] = "typeof"
  1251		       b[3] = "geo:Point"
  1252		       S = S _mrwresponse(m,div,"","","","","","",a,b,1,"!")
  1253		    }
  1254		    else {
  1255		       a[1] = "xmlns:geo"
  1256		       b[1] = "http://www.w3.org/2003/01/geo/wgs84_pos#"
  1257		       a[2] = "rel"
  1258		       b[2] = "geo:Point"
  1259	     	       if (subrel)
  1260			  S = S _mrwresponse(m1,"","span","","","","","",a,b,1,"!")
  1261		       else
  1262			  S = S _mrwresponse(m,"","span","","","","","",a,b,1,"!")
  1263		    }
  1264	
  1265		    # Ignore any "name" and "schema" spects of the (:v-geo ...:)
  1266		    # CPI here, as they are not part of the RDFa specs and I
  1267		    # dont't know how to code them. It will always be possible
  1268		    # to support them in the future by inserting the appropriate
  1269		    # code here.
  1270		 }
  1271	         else if (name ~ /^lat(itude)?$/) {
  1272	     	    a[1] = "property"
  1273	     	    b[1] = "geo:lat"
  1274		    if (subrel)
  1275			S = S _mrwresponse(m1,"","","","","","span",value,a,b,1,"!",1)
  1276		    else
  1277			S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1278	         }
  1279	         else if (name ~ /^long(itude)?$/) {
  1280	     	    a[1] = "property"
  1281	     	    b[1] = "geo:long"
  1282		    if (subrel)
  1283			S = S _mrwresponse(m1,"","","","","","span",value,a,b,1,"!",1)
  1284		    else
  1285			S = S _mrwresponse(m,"","","","","","span",value,a,b,1,"!",1)
  1286	         }
  1287		 else {
  1288		    scantype = oldtype[i--]
  1289		    s = saved s				# no match, stack it back.
  1290		    flag = 1
  1291		 }
  1292	      }
  1293	
  1294	      else S = S "(:v-" name " " value ":)"
  1295	  }
  1296	
  1297	  # Close the resulting top-level structure. Leave any subrel(s) alone
  1298	  # for the moment, but it may be necessary to try and uncomment next
  1299	  # line if any structure errors occur under certain circumstances.
  1300	
  1301	  #if (subrel) S = S _mrwresponse(m1,".","","","","","","",a,b,1,"!")
  1302	
  1303	  S = S _mrwresponse(m,".","","","","","","",a,b,1,"!")
  1304	
  1305	  # Better comment this out while debugging.
  1306	  S = S _RRMATCH
  1307	
  1308	  gsub(/\001/,":)",S)
  1309	
  1310	  return S
  1311	}
  1312