Bugs and <pb> handling

oeuvres · Sep 26, 2024 · 64cd43d · 64cd43d
1 parent a312672
commit 64cd43d
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 6 deletions.
diff --git a/docx/divs.xsl b/docx/divs.xsl
@@ -108,6 +108,11 @@
 
   <xsl:template match="tei:item/@level"/>
 
+  <xsl:template match="tei:pb">
+    <pb>
+      <xsl:value-of select="normalize-space(.)"/>
+    </pb>
+  </xsl:template>
 
   <xsl:template match="tei:head[@level]">
     <xsl:choose>

diff --git a/docx/docx_teilike.xsl b/docx/docx_teilike.xsl
@@ -445,13 +445,13 @@ Seen
       <xsl:variable name="val" select="w:rPr/w:u/@w:val"/>
       <xsl:choose>
         <xsl:when test="not(w:rPr/w:u)"/>
-        <xsl:when test="not($w:style) and not($w:style/w:rPr/w:u)"/>
         <xsl:when test="$val = '0' or $val='false' or $val = 'off' or $val = 'none'"/>
         <xsl:when test="$val != ''">
           <xsl:value-of select="$val"/>
         </xsl:when>
       </xsl:choose>
     </xsl:variable>
+
     <!-- small caps -->
     <xsl:variable name="sc">
       <xsl:variable name="val" select="w:rPr/w:smallCaps/@w:val"/>
@@ -479,7 +479,7 @@ Seen
         <xsl:when test="w:rPr/w:i">i</xsl:when>
       </xsl:choose>
     </xsl:variable>
-    <!-- bold, dangerous in titles, to think -->
+    <!-- bold -->
     <xsl:variable name="bval" select="string(w:rPr/w:b/@w:val)"/>
     <xsl:variable name="bCsval" select="string(w:rPr/w:bCs/@w:val)"/>
     <xsl:variable name="b">

diff --git a/docx/teilike_pcre.tsv b/docx/teilike_pcre.tsv
@@ -9,16 +9,27 @@ search	replace	?comment
 
 # variables ends with the first replacement rule
 
+# Page breaks
+\[([pf]\.[  ][^<>\]]+)\]	<pb>\1</pb>	Page break with no style
+</pb><pb>		broken pb
+<pb>([  \n]*)</pb>	 	normalize space 
+<([a-z]+)>(<pb>[^<>]+</pb>)</\1>	\2	false para, doubles <pb> upper, inlines
+<([a-z]+)>(<pb>[^<>]+</pb>)</\1>	\2	…
+<([a-z]+)>(<pb>[^<>]+</pb>)</\1>	\2	…
+
 # SECTIONS
 
+(<pb>[^<]+</pb>)\s*((<\?div( /)?\?>\s*)+)	\2\n\1	first page break of a section
 <\?(div|list)\?>	<\1>	generated hierarchy
 <\?(div|list) /\?>	</\1>	generated hierarchy
-\n* *</($parent)>\s*<($parent)[^>]*> *\n*	\n\n
+\n* *</($parent)>\s*<\1[^>]*> *\n*	\n\n
+
+(<head($atts)>)\s+	\1	trim spaces
+(<head($atts))>@	\1 type="chapter">	<head>@chapter
 
 -[\*]\n	-\n	odd chars from OCR
 
-<pb>(\s*)</pb>	\1
-</pb><pb>		broken pb
+
 
 <($inline)($atts)>(\s*)(<lb($atts)/>)(\s*)</\1>	\3\4\6	inline around line break
 </($inline)>( *)<\1($atts)>	\2	linked inline, block tags separed by \n
@@ -30,7 +41,7 @@ search	replace	?comment
 (\s+)</($inline)>	</\2>\1	spaces at end of inline, put after
 <($inline)($atts)>($pun)</\1>	\3	inline without word char
 # ([,: \s])(</($inline)>)	\2\1	soft pun at end,</inline> 
-<($inline)($atts)></($inline)>		empty inlines
+<($inline)($atts)></\1>		empty inlines
 <($inline)($atts)/>		empty inlines