improved rclid3 and rclogg

2008-10-08 08:27:34 +00:00 · 2008-10-08 08:27:34 +00:00 · 49388c8748
commit 49388c8748
parent 954de37067
12 changed files with 79 additions and 101 deletions
--- a/src/filters/rclabw
+++ b/src/filters/rclabw
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rclabw,v 1.2 2007-06-15 11:41:50 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: rclabw,v 1.3 2008-10-08 08:27:34 dockes Exp $  (C) 2004 J.F.Dockes
 # Parts taken from Estraier:
 #================================================================
 # Estraier: a personal full-text search system
@ -83,13 +83,6 @@ umask 77

 checkcmds iconv sed

-# check the input file existence
-if test ! -f "$infile"
-then
-  printf '%s: %s: no such file\n' "$progname" "$infile"
-  exit 1
-fi
-
 encoding=`sed -e  '/<?xml version=/s/"?>$//' \
 	      -e '/^<?xml version=/s/.*encoding="//p;D;q' \
 	      -e D \
--- a/src/filters/rcldjvu
+++ b/src/filters/rcldjvu
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rcldjvu,v 1.5 2007-06-08 13:51:08 dockes Exp $  (C) 2005 J.F.Dockes
+# @(#$Id: rcldjvu,v 1.6 2008-10-08 08:27:34 dockes Exp $  (C) 2005 J.F.Dockes

 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
@ -103,13 +103,6 @@ umask 77

 checkcmds djvutxt djvused awk

-# check the input file existence
-if test ! -f "$infile"
-then
-  printf '%s: %s: no such file\n' "$progname" "$infile"
-  exit 1
-fi
-
 # Title: we try to extract it from the annotations. djvused outputs string
 # in C/awk \-escaped notation. Awk can only process this in string
 # constants, so we have a first awk pass to create an awk program to parse
--- a/src/filters/rclid3
+++ b/src/filters/rclid3
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rclid3,v 1.2 2007-11-09 11:54:59 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: rclid3,v 1.3 2008-10-08 08:27:34 dockes Exp $  (C) 2004 J.F.Dockes
 #================================================================
 # rclid3
 # Handle audio files for recoll. This uses id3info to extract tags
@ -83,25 +83,52 @@ checkcmds id3info
 echo '<html><head>'
 #echo '<title>' "$title" '</title>'
 echo '<meta http-equiv="Content-Type" content="text/html;charset=UTF-8">'
-echo '</head><body>'
-echo '<pre>'

 nocaptionexp='s/===.*://'
-kpcaptionexp='s/===[^(]*//'

 if test X$RECOLL_FILTER_FORPREVIEW = Xyes ; then
-    captionexp=$kpcaptionexp
-else
-    captionexp=$nocaptionexp
-fi
-
+echo '</head><body><pre>'
 id3info "$infile" | \
   sed -e '/Tag information for/d' \
-       -e "$captionexp" \
-       -e 's/</&lt;/g' -e 's/&/&amp;/g' \
+       -e 's/===[^(]*//' \
+       -e 's/</&lt;/g' -e 's/&/&amp;/g' -e 's/"/&quot;/g' -e 's/>/&gt;/g' 
+echo '</pre></body></html>'

-echo '</pre>'
-echo '</body></html>'
+else
+# Indexing. There are many other fields which we could process this way.
+# Would also need that some fields are set up in the "fields" config
+# file to be really useful
+id3info "$infile" | awk -F: \
+'{
+   value=""
+   for (i = 2; i <=NF; i++) {
+     value = value " " $i
+   }
+   sub("&", "&amp;", value)
+   sub("\"", "&quot;", value)
+   sub("<", "&lt;", value)
+   sub(">", "&gt;", value)
+}
+/TIT2/{
+   printf "<meta name=\"title\" content=\"%s\">\n", value
+   body=body value "\n"
+   next
+}
+/TPE1/{
+   printf "<meta name=\"artist\" content=\"%s\">\n", value
+   body=body value "\n"
+   next
+}
+/TALB/{
+   printf "<meta name=\"album\" content=\"%s\">\n", value
+   body=body value "\n"
+   next
+}
+{
+   body=body value "\n"
+}
+END{print "</head><body><pre>";print body;print "</pre></body></html>"}
+'
+fi

-# exit normally
 exit 0
--- a/src/filters/rclkwd
+++ b/src/filters/rclkwd
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rclkwd,v 1.1 2007-06-08 14:01:30 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: rclkwd,v 1.2 2008-10-08 08:27:34 dockes Exp $  (C) 2004 J.F.Dockes
 # Parts taken from Estraier:
 #================================================================
 # Estraier: a personal full-text search system
@ -87,13 +87,6 @@ umask 77

 checkcmds awk unzip gunzip tar

-# check the input file existence
-if test ! -f "$infile"
-then
-  printf '%s: %s: no such file\n' "$progname" "$infile"
-  exit 1
-fi
-
 # We need a temporary directory
 if test z"$RECOLL_TMPDIR" != z; then
   ttdir=$RECOLL_TMPDIR
--- a/src/filters/rclogg
+++ b/src/filters/rclogg
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rclogg,v 1.1 2007-10-02 14:00:47 dockes Exp $  (C) 2007 J.F.Dockes
+# @(#$Id: rclogg,v 1.2 2008-10-08 08:27:34 dockes Exp $  (C) 2007 J.F.Dockes
 #================================================================
 # rclogg
 # Handle ogg audio files for recoll.
@ -74,33 +74,37 @@ fi
 # !! Leave the following line unmodified !
 #ENDRECFILTCOMMONCODE

-checkcmds ogginfo
+checkcmds ogginfo sed awk

 # output the result
 echo '<html><head>'
-#echo '<title>' "$title" '</title>'
 echo '<meta http-equiv="Content-Type" content="text/html;charset=UTF-8">'
-echo '</head><body>'
-echo '<pre>'

-nocaptionexp='/=/s/.*=//p'
-kpcaptionexp='/=/p'
+if test X$RECOLL_FILTER_FORPREVIEW = Xyes 
+then
+
+echo '</head><body><pre>'
+ogginfo "$infile" | grep '=' | sed -e 's/&/&amp;/g' -e 's/"/&quot;/g' \
+    -e 's/</&lt;/g' -e 's/>/&gt;/g'
+echo '</pre></body></html>'

-if test X$RECOLL_FILTER_FORPREVIEW = Xyes ; then
-    captionexp=$kpcaptionexp
 else
-    captionexp=$nocaptionexp
+    ogginfo "$infile" | grep '=' | awk -F= \
+      '{sub("^[ 	]+", "", $1)  
+        sub("&", "&amp;", $2)
+        sub("\"", "&quot;", $2)
+        sub("<", "&lt;", $2)
+        sub(">", "&gt;", $2)
+        printf "<meta name=\"%s\" content=\"%s\">\n", $1, $2
+        body=body $2 "\n"
+       }
+    END{print "</head><body><pre>";print body;print "</pre></body></html>"}'
 fi

 # Note: ogginfo output is unparseable: the COMMENT field can be
 # multiline, but there is no way to detect the end of the 'User
 # comments' area
-ogginfo "$infile" | \
-   sed -n \
-    -e "$captionexp"

-echo '</pre>'
-echo '</body></html>'

 # exit normally
 exit 0
--- a/src/filters/rclopxml
+++ b/src/filters/rclopxml
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rclopxml,v 1.2 2008-09-01 17:31:47 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: rclopxml,v 1.3 2008-10-08 08:27:34 dockes Exp $  (C) 2004 J.F.Dockes
 #================================================================
 # rcldocx
 # Extract text from an openxml msword file (will be extended for spreadsheets)
@ -80,13 +80,6 @@ umask 77

 checkcmds xsltproc unzip

-# check the input file existence
-if test ! -f "$infile"
-then
-  printf '%s: %s: no such file\n' "$progname" "$infile"
-  exit 1
-fi
-
 # We need a temporary directory
 if test z"$RECOLL_TMPDIR" != z; then
   ttdir=$RECOLL_TMPDIR
--- a/src/filters/rclppt
+++ b/src/filters/rclppt
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rclppt,v 1.3 2007-06-08 13:51:09 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: rclppt,v 1.4 2008-10-08 08:27:34 dockes Exp $  (C) 2004 J.F.Dockes
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 2 of the License, or
@ -100,10 +100,6 @@ umask 77
 #ENDRECFILTCOMMONCODE

 checkcmds catppt
-if test X$cmdsok = X0 ; then
-   printf "Catppt not found"
-   exit 1
-fi

 # output the result
 echo '<html><head>'
--- a/src/filters/rclsoff
+++ b/src/filters/rclsoff
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rclsoff,v 1.11 2007-06-15 11:41:50 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: rclsoff,v 1.12 2008-10-08 08:27:34 dockes Exp $  (C) 2004 J.F.Dockes
 # Parts taken from Estraier:
 #================================================================
 # Estraier: a personal full-text search system
@ -89,13 +89,6 @@ umask 77

 checkcmds awk iconv unzip

-# check the input file existence
-if test ! -f "$infile"
-then
-  printf '%s: %s: no such file\n' "$progname" "$infile"
-  exit 1
-fi
-
 # We need a temporary directory
 if test z"$RECOLL_TMPDIR" != z; then
   ttdir=$RECOLL_TMPDIR
--- a/src/filters/rclsvg
+++ b/src/filters/rclsvg
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rclsvg,v 1.2 2008-02-03 16:05:57 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: rclsvg,v 1.3 2008-10-08 08:27:34 dockes Exp $  (C) 2004 J.F.Dockes
 # Parts taken from Estraier:
 #================================================================
 # Estraier: a personal full-text search system
@ -83,13 +83,6 @@ umask 77

 checkcmds iconv sed

-# check the input file existence
-if test ! -f "$infile"
-then
-  printf '%s: %s: no such file\n' "$progname" "$infile"
-  exit 1
-fi
-
 encoding=`sed -ne '/<?xml/s/.*encoding="\([^"]*\).*/\1/p' < $infile`

 if test X$encoding = X ; then encoding=UTF-8;fi
--- a/src/filters/rclxls
+++ b/src/filters/rclxls
@ -1,5 +1,5 @@
 #!/bin/sh
-# @(#$Id: rclxls,v 1.4 2007-06-08 13:51:09 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: rclxls,v 1.5 2008-10-08 08:27:34 dockes Exp $  (C) 2004 J.F.Dockes
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 # the Free Software Foundation; either version 2 of the License, or
@ -100,17 +100,7 @@ umask 77
 # !! Leave the following line unmodified !
 #ENDRECFILTCOMMONCODE

-# check the input file existence
-if test ! -f "$infile"
-then
-  printf '%s: %s: no such file\n' "$progname" "$infile"
-  exit 1
-fi
 checkcmds xls2csv
-if test X$cmdsok = X0 ; then
-   printf "xls2csv not found"
-   exit 1
-fi

 # output the result
 echo '<html><head>'
--- a/src/sampleconf/fields
+++ b/src/sampleconf/fields
@ -1,4 +1,4 @@
-# @(#$Id: fields,v 1.4 2008-10-07 06:44:23 dockes Exp $  (C) 2007 J.F.Dockes
+# @(#$Id: fields,v 1.5 2008-10-08 08:27:34 dockes Exp $  (C) 2007 J.F.Dockes
 # Field names configuration. This defines how one may search ie for 
 # author:Hemingway
 # Important: 
@ -53,7 +53,7 @@ stored = author

 [aliases]
 abstract = summary dc:summary description xesam:description
-author = creator dc:creator xesam:author xesam:creator
+author = creator dc:creator xesam:author xesam:creator from
 caption = title title dc:title subject
 # catg = dc:type contentCategory
 dbytes = size xesam:size
--- a/src/sampleconf/mimeconf
+++ b/src/sampleconf/mimeconf
@ -1,4 +1,4 @@
-# @(#$Id: mimeconf,v 1.44 2008-09-28 14:20:50 dockes Exp $  (C) 2004 J.F.Dockes
+# @(#$Id: mimeconf,v 1.45 2008-10-08 08:27:34 dockes Exp $  (C) 2004 J.F.Dockes

 # Recoll : associations of mime types to processing filters.
 # There are different sections for decompression, 'interning' for indexing
@ -26,12 +26,15 @@ application/x-bzip2 =  uncompress rcluncomp bunzip2 %f %t
 # The external "exec" filters are typically scripts. They output the
 # document in simple html format, have a look at the scripts.
 [index]
-application/msword = exec rcldoc
+# Note: rcldoc did some work to splice hyphenated words at eol. Seems
+# actually not needed because antiword apparently does it too
+# application/msword = exec rcldoc
+application/msword = exec antiword -t -i 1 -m UTF-8;mimetype=text/plain;charset=utf-8
 application/ogg = exec rclogg
 application/pdf = exec rclpdf
 application/postscript = exec rclps
-application/vnd.ms-excel = exec rclxls
-application/vnd.ms-powerpoint = exec rclppt
+application/vnd.ms-excel = exec xls2csv -c "	" -d utf-8;charset=utf-8;mimetype=text/plain
+application/vnd.ms-powerpoint = exec catppt -d utf-8;charset=utf-8;mimetype=text/plain
 application/vnd.openxmlformats-officedocument.wordprocessingml.document = \
 exec rclopxml
 application/vnd.openxmlformats-officedocument.wordprocessingml.template = \
@ -54,7 +57,7 @@ application/vnd.sun.xml.math = exec rclsoff
 application/vnd.sun.xml.writer = exec rclsoff
 application/vnd.sun.xml.writer.global = exec rclsoff
 application/vnd.sun.xml.writer.template = exec rclsoff
-application/vnd.wordperfect = exec rclwpd
+application/vnd.wordperfect = exec wpd2html;mimetype=text/html
 application/x-abiword = exec rclabw
 application/x-dvi = exec rcldvi
 application/x-flac = exec rclflac
@ -72,7 +75,7 @@ image/svg+xml = exec rclsvg
 message/rfc822 = internal
 text/html  = internal 
 text/plain = internal 
-text/rtf = exec rclrtf
+text/rtf = exec unrtf --nopict --html;charset=iso-8859-1;mimetype=text/html
 text/x-c = exec rcltext
 text/x-gaim-log = exec rclgaim
 text/x-html-sidux-man = exec rclsiduxman