Add epub search helper (#2867)

linuxmint · Jun 10, 2022 · d5644ad · d5644ad
1 parent 1140fcf
commit d5644ad
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 0 deletions.
diff --git a/search-helpers/epub2text.nemo_search_helper b/search-helpers/epub2text.nemo_search_helper
@@ -0,0 +1,5 @@
+[Nemo Search Helper]
+TryExec=unzip;html2text;nemo-epub2text;
+Exec=nemo-epub2text %s
+MimeType=application/epub+zip;
+Priority=100
diff --git a/search-helpers/nemo-epub2text b/search-helpers/nemo-epub2text
@@ -0,0 +1,14 @@
+#!/usr/bin/env bash
+
+# Requires unzip package
+# ==================================
+# 1. Get a list of all xhtml/html/htm files, exclude titlepage.xhtml (if present)
+#    It appears that the zipped files _never_ contain problem characters such as spaces...
+# 2. Extract the html files and convert to text (UTF-8 output is available).
+# ==================================
+
+# 1. Get a list of xhtml/html/htm files [using unzip's weird regular expression] - and exclude any named titlepage/toc/copyright
+files=$(unzip -Z1 "$1" \*.*htm* | egrep -v 'titlepage.*|toc.*|copyright.*')
+
+# 2. Uncompress each of the files and process with html2text.
+unzip -cqq "$1" $files | html2text -o -