“Support” X·M·L 1.1

The X·M·L 1.1 “support” amounts to deleting the declaration and
replacing any character escapes for C·0 controls with
`U+0091 PRIVATE USE ONE`, which is a valid character in X·M·L 1.0.
This is done entirely in `sed`, so it’s not perfect, but it should be
“good enough”.

diff --git a/GNUmakefile b/GNUmakefile
index 16979033e5bbed8ae3f878629bb705f482792379..c953ac738ef80071cc89466df708c60707c72e86 100644 (file)
--- a/GNUmakefile
+++ b/GNUmakefile
@@ -386,6 +386,13 @@ override id = $(XMLLINT) --xpath '/*/*[local-name()="id" and namespace-uri()="ur
 # (callable) Sanitize and wrap the provided plaintext file in X·M·L, printing to `stdout´.
 override wrapplaintext = $(PRINTF) '%s\n' "$$($(PRINTF) '%b' '<?xml version=\042\061.0\042?>\n<script xmlns=\042http://www.w3.org/1999/xhtml\042 type=\042$(call typeoffile,$1)\042><![CDATA['; $(CAT) $(call quote,$1) | $(TR) '\000\013\014' '\032\011\012' | $(SED) $$($(PRINTF) '%s%b' 's/]]>/]]]]><!\[CDATA\[>/g;s/\xEF\xBF\xBE/�/g;s/\xEF\xBF\xBF/�/g;$$!s/\r$$//g;s/\r/\n/g;$$!s/\xC2\x85$$//g;s/\xC2\x85/\n/g;s/\xE2\x80\xA8/\n/g;' 's/[\0001-\0010]/�/g;s/[\0016-\0037]/�/g'); $(PRINTF) '%s' ']]></script>')"
 
 # (callable) Sanitize and wrap the provided plaintext file in X·M·L, printing to `stdout´.
 override wrapplaintext = $(PRINTF) '%s\n' "$$($(PRINTF) '%b' '<?xml version=\042\061.0\042?>\n<script xmlns=\042http://www.w3.org/1999/xhtml\042 type=\042$(call typeoffile,$1)\042><![CDATA['; $(CAT) $(call quote,$1) | $(TR) '\000\013\014' '\032\011\012' | $(SED) $$($(PRINTF) '%s%b' 's/]]>/]]]]><!\[CDATA\[>/g;s/\xEF\xBF\xBE/�/g;s/\xEF\xBF\xBF/�/g;$$!s/\r$$//g;s/\r/\n/g;$$!s/\xC2\x85$$//g;s/\xC2\x85/\n/g;s/\xE2\x80\xA8/\n/g;' 's/[\0001-\0010]/�/g;s/[\0016-\0037]/�/g'); $(PRINTF) '%s' ']]></script>')"
 

+# (callable) Check if the provided X·M·L file is X·M·L 1.1, and if so, coerce to X·M·L 1.0 as best as possible, printing the result (or the original file contents) to `stdout´.
+#
+# The X·M·L declaration will be dropped and character escapes for C0 control codes will be replaced with a literal `U+0091 PRIVATE USE ONE´, which is invalid in X·M·L 1.1, but valid X·M·L 1.0 (making the replacement obvious).
+#
+# This isn’t a perfect substitution (it makes some assumptions about the format of the underlying X·M·L), but it should be workable for most sensible, welformed files.
+override serializexml = $(SED) "$$($(PRINTF) '%b' '/<?xml[ \t]\\{1,\\}version=[\0042\0047]1.1/,$${ s/<?xml[^>]*?>/<!--<?xml version=\00421.1\0042?>-->/\n s/&\0043x0*[1-8BCEFbcef];/\0302\0221/g\n s/&\0043x0*1[0-9A-Fa-f];/\0302\0221/g\n s/&\00430*[1-8];/\0302\0221/g\n s/&\00430*1[124-9];/\0302\0221/g\n s/&\00430*2[0-9];/\0302\0221/g\n s/&\00430*3[01];/\0302\0221/g\n}')" < $(call quote,$1) | $(SED) "$$(PRINTF '%b' ':a\n/^\\n*$$/{ $$d\n N\n ba\n}')"
+

 # ─ ¶ Phony Targets ───────────────────────────────────────────────────
 
 # Compile all files, or error if any are recursive.
 # ─ ¶ Phony Targets ───────────────────────────────────────────────────
 
 # Compile all files, or error if any are recursive.


@@ -497,7 +504,7 @@ $(BUILDDIR)/parser.xslt : $(BUILDDIR)/parser.catalog $(THISDIR)/lib/catalog2pars
 $(call parsed,$(sourcefiles) $(sourceincludes)) : % : $$(call unparsed,$$@) $(BUILDDIR)/parser.xslt $(typeupdates)
        @$(PRINTF) '%s\n' $(call quote,Processing `$<´…)
        $(silent)$(call ensuredirectory,$(dir $@))
 $(call parsed,$(sourcefiles) $(sourceincludes)) : % : $$(call unparsed,$$@) $(BUILDDIR)/parser.xslt $(typeupdates)
        @$(PRINTF) '%s\n' $(call quote,Processing `$<´…)
        $(silent)$(call ensuredirectory,$(dir $@))

-       $(silent)$(if $(filter $<,$(assetfiles)),$(PRINTF) '%s\n' $(call quote,<object xmlns="http://www.w3.org/1999/xhtml" type="$(call typeoffile,$<)" data="$(call datauri,$<)"/>) > $(call quote,$@),$(if $(filter $<,$(plaintextfiles)),$(call wrapplaintext,$<),$(CAT) $(call quote,$<)) | $(XSLTPROC) -o $(call quote,$@) --stringparam BUILDTIME $$(TZ= $(DATE) '+%Y-%m-%dT%H:%M:%SZ') --stringparam IDENTIFIER $(call quote,$(call localuri,$<)) --stringparam SRCTIME $$(TZ= $(STAT) -f '%Sm' -t '%Y-%m-%dT%H:%M:%SZ' $(call quote,$<)) --stringparam CKSUM $$($(CKSUM) $(call quote,$<) | $(SED) 's/[ ].*//')$(if $(THISREV), --stringparam THISREV $(call quote,$(THISREV)),)$(if $(SRCREV), --stringparam SRCREV $(call quote,$(SRCREV)),) $(call quote,$(BUILDDIR)/parser.xslt) -)
+       $(silent)$(if $(filter $<,$(assetfiles)),$(PRINTF) '%s\n' $(call quote,<object xmlns="http://www.w3.org/1999/xhtml" type="$(call typeoffile,$<)" data="$(call datauri,$<)"/>) > $(call quote,$@),$(if $(filter $<,$(plaintextfiles)),$(call wrapplaintext,$<),$(call serializexml,$<)) | $(XSLTPROC) -o $(call quote,$@) --stringparam BUILDTIME $$(TZ= $(DATE) '+%Y-%m-%dT%H:%M:%SZ') --stringparam IDENTIFIER $(call quote,$(call localuri,$<)) --stringparam SRCTIME $$(TZ= $(STAT) -f '%Sm' -t '%Y-%m-%dT%H:%M:%SZ' $(call quote,$<)) --stringparam CKSUM $$($(CKSUM) $(call quote,$<) | $(SED) 's/[ ].*//')$(if $(THISREV), --stringparam THISREV $(call quote,$(THISREV)),)$(if $(SRCREV), --stringparam SRCREV $(call quote,$(SRCREV)),) $(call quote,$(BUILDDIR)/parser.xslt) -)

 
 # Generate a catalog of all parsed files, for use when processing includes.
 #
 
 # Generate a catalog of all parsed files, for use when processing includes.
 #