Merge branch 'links'

simsapa · Nov 17, 2019 · bb97d7c · bb97d7c
2 parents a174112 + 5db6acf
commit bb97d7c
Show file tree

Hide file tree

Showing 33 changed files with 2,683 additions and 439 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -1,6 +1,6 @@
 [package]
 name = "simsapa_dictionary"
-version = "0.1.0"
+version = "0.2.0-alpha.1"
 authors = ["Gambhiro <gambhiro.bhikkhu.85@gmail.com>"]
 edition = "2018"
 license = "MIT"

diff --git a/README.md b/README.md
@@ -75,8 +75,6 @@ ndped-example.md
     book_id = "NcpedDictionarySimsapa"
     created_date_human = ""
     created_date_opf = ""
-    is_epub = true
-    is_mobi = false
     ```
 
     --- DICTIONARY WORD ENTRIES ---
@@ -162,7 +160,7 @@ SUBCOMMANDS:
 Process a Markdown file and generate an EPUB or MOBI dictionary.
 
 USAGE:
-    simsapa_dictionary markdown_to_ebook [FLAGS] [OPTIONS] --ebook_format <FORMAT>
+    simsapa_dictionary markdown_to_ebook [FLAGS] [OPTIONS] --output_format <FORMAT>
 
 FLAGS:
         --dont_remove_generated_files    Turns off the removal of the generated OPF, HTML, etc. files used to create the
@@ -176,7 +174,7 @@ FLAGS:
 
 OPTIONS:
         --dict_label <LABEL>            Use this dict_label property, instead of the one defined in the Markdown file.
-        --ebook_format <FORMAT>         Either EPUB or MOBI. [default: EPUB]  [possible values: EPUB, Epub, epub, MOBI,
+        --output_format <FORMAT>        Either EPUB or MOBI. [default: EPUB]  [possible values: EPUB, Epub, epub, MOBI,
                                         Mobi, mobi]
         --kindlegen_path <PATH>         The KindleGen tool must be available either (a) in the current folder with this
                                         tool, (b) in the system PATH, (c) declared with this option.

diff --git a/appveyor.yml b/appveyor.yml
@@ -1,4 +1,4 @@
-version: "0.1.0.{branch}.{build}"
+version: "0.2.0-alpha.1.{branch}.{build}"
 
 image: Visual Studio 2019
 

diff --git a/assets/OEBPS/entries-epub.xhtml b/assets/OEBPS/entries-epub.xhtml
@@ -6,32 +6,15 @@
 
 {{#each group.dict_words}}
 <h4>{{#if ../meta.word_prefix}}{{../meta.word_prefix}} {{/if}}{{word_header.word}}</h4>
-{{grammar_phonetic_transliteration word_header ../meta.use_velthuis ~}}
 {{#if word_header.dict_label ~}}
 <p>[{{word_header.dict_label}}]</p>
 {{/if ~}}
-{{#if word_header.summary ~}}
-<p>{{word_header.summary}}</p>
-{{/if ~}}
+{{grammar_phonetic_transliteration word_header ../meta.add_velthuis ~}}
 
+{{word_list "Also written as:" word_header.also_written_as ~}}
 {{markdown definition_md}}
-
-{{#if word_header.synonyms}}
-<p>Synonyms:</p>
-<ul>
-{{#each word_header.synonyms}}
-<li>{{this}}</li>
-{{/each}}
-</ul>
-{{/if}}
-
-{{#if word_header.antonyms}}
-<p>Antonyms:</p>
-<ul>
-{{#each word_header.antonyms}}
-<li>{{this}}</li>
-{{/each}}
-</ul>
-{{/if}}
+{{word_list "Synonyms:" word_header.synonyms ~}}
+{{word_list "Antonyms:" word_header.antonyms ~}}
+{{word_list "See also:" word_header.see_also ~}}
 
 {{/each}}
diff --git a/assets/OEBPS/entries-mobi.xhtml b/assets/OEBPS/entries-mobi.xhtml
@@ -7,26 +7,14 @@
 {{! Spaces and newlines in <idx:entry> are significant for dictionary entry search}}
 
 {{#each group.dict_words}}
-<idx:entry><idx:orth><h4>{{word_header.word}}</h4>{{grammar_phonetic_transliteration word_header ../meta.use_velthuis ~}}<p>{{#if word_header.dict_label}} <i>{{word_header.dict_label}}</i>{{/if}}{{#if word_header.summary}} <i>{{word_header.summary}}</i>{{/if}}{{#if word_header.inflections}}</p><idx:infl>{{#each word_header.inflections}} <idx:iform value="{{this}}"/>{{/each}}</idx:infl>{{/if}}</idx:orth></idx:entry>
+<idx:entry><idx:orth><h4>{{word_header.word}}</h4><p>{{#if word_header.dict_label}} <i>{{word_header.dict_label}}</i>{{/if}}{{#if word_header.summary}} <i>{{word_header.summary}}</i>{{/if}}{{#if word_header.inflections}}</p><idx:infl>{{#each word_header.inflections}} <idx:iform value="{{this}}"/>{{/each}}</idx:infl>{{/if}}</idx:orth></idx:entry>
 
-{{markdown definition_md}}
-
-{{#if word_header.synonyms}}
-<p>Synonyms:</p>
-<ul>
-{{#each word_header.synonyms}}
-<li>{{this}}</li>
-{{/each}}
-</ul>
-{{/if}}
+{{grammar_phonetic_transliteration word_header ../meta.add_velthuis ~}}
 
-{{#if word_header.antonyms}}
-<p>Antonyms:</p>
-<ul>
-{{#each word_header.antonyms}}
-<li>{{this}}</li>
-{{/each}}
-</ul>
-{{/if}}
+{{word_list "Also written as:" word_header.also_written_as ~}}
+{{markdown definition_md}}
+{{word_list "Synonyms:" word_header.synonyms ~}}
+{{word_list "Antonyms:" word_header.antonyms ~}}
+{{word_list "See also:" word_header.see_also ~}}
 
 {{/each}}
diff --git a/assets/OEBPS/package.opf b/assets/OEBPS/package.opf
@@ -13,7 +13,7 @@
 
     <meta name="cover" content="item_cover"></meta>
 
-    {{#if meta.is_mobi ~}}
+    {{#if (eq output_format "Mobi") ~}}
     <x-metadata>
       <output encoding="utf-8"></output>
       <EmbeddedCover>{{meta.cover_path}}</EmbeddedCover>
@@ -27,7 +27,7 @@
   <manifest>
     <item id="htmltoc" href="htmltoc.xhtml" media-type="application/xhtml+xml" properties="nav"/>
 
-    {{#if meta.is_epub ~}}
+    {{#if (eq output_format "Epub") ~}}
     <item id="cover_page" href="cover.xhtml" media-type="application/xhtml+xml" properties="svg"/>
     {{/if ~}}
 
@@ -48,9 +48,10 @@
   </manifest>
 
   <spine toc="ncx">
-    {{#if meta.is_epub ~}}
+    {{#if (eq output_format "Epub") ~}}
     <itemref idref="cover_page"/>
-    {{else}}
+    {{/if ~}}
+    {{#if (eq output_format "Mobi") ~}}
     <itemref idref="item_cover"/>
     {{/if ~}}
 

diff --git a/assets/simsapa_md_to_epub.bat b/assets/simsapa_md_to_epub.bat
@@ -11,7 +11,7 @@ SET DEST_FILE=%SRC_NAME%.epub
 simsapa_dictionary.exe markdown_to_ebook ^
     --source_path "%~1" ^
     --dict_label "" ^
-    --ebook_format epub ^
+    --output_format epub ^
     --output_path "%DEST_FILE%"
 
 Exit /b

diff --git a/assets/simsapa_md_to_epub.sh b/assets/simsapa_md_to_epub.sh
@@ -7,6 +7,6 @@ DEST_FILE="$SRC_NAME.epub"
 simsapa_dictionary markdown_to_ebook \
     --source_path "$1" \
     --dict_label "" \
-    --ebook_format epub \
+    --output_format epub \
     --output_path "$DEST_FILE"
 
diff --git a/assets/simsapa_md_to_mobi.bat b/assets/simsapa_md_to_mobi.bat
@@ -11,7 +11,7 @@ SET DEST_FILE=%SRC_NAME%.mobi
 simsapa_dictionary.exe markdown_to_ebook ^
     --source_path "%~1" ^
     --dict_label "" ^
-    --ebook_format mobi ^
+    --output_format mobi ^
     --output_path "%DEST_FILE%" ^
     --mobi_compression 0 ^
     --kindlegen_path kindlegen.exe

diff --git a/assets/simsapa_md_to_mobi.sh b/assets/simsapa_md_to_mobi.sh
@@ -9,7 +9,7 @@ DEST_FILE="$SRC_NAME.mobi"
 simsapa_dictionary markdown_to_ebook \
     --source_path "$1" \
     --dict_label "" \
-    --ebook_format mobi \
+    --output_format mobi \
     --output_path "$DEST_FILE" \
     --mobi_compression 0
 
diff --git a/assets/stardict_textual.xml b/assets/stardict_textual.xml
@@ -25,7 +25,11 @@
 
       <definition type="h">
         <![CDATA[
-{{grammar_phonetic_transliteration word_header ../meta.use_velthuis ~}}
+{{#if word_header.dict_label ~}}
+<p>[{{word_header.dict_label}}]</p>
+{{/if ~}}
+{{grammar_phonetic_transliteration word_header ../meta.add_velthuis ~}}
+{{word_list "Also written as:" word_header.also_written_as ~}}
 {{markdown definition_md ~}}
 {{word_list "Synonyms:" word_header.synonyms ~}}
 {{word_list "Antonyms:" word_header.antonyms ~}}

diff --git a/scripts/build_dicts_for_release.sh b/scripts/build_dicts_for_release.sh
@@ -3,41 +3,89 @@
 SRC_DIR=../simsapa-dictionary-data
 OUT_DIR=../simsapa-dictionary_releases/new-release
 
+if [ -z "$OUT_DIR" ]; then
+    rm -r "$OUT_DIR"
+fi
+
+mkdir -p "$OUT_DIR"
+
+PROJ_ROOT=$(pwd)
+
 KINDLEGEN_PATH="$HOME/lib/kindlegen/kindlegen"
 
-# === Combined ===
+STARDICT_TEXT2BIN="/usr/lib/stardict-tools/stardict-text2bin"
+
+# === Individual ===
 
 for i in dhammika dppn ncped nyana pts; do
+    cd "$PROJ_ROOT"
 
     cargo run -- markdown_to_ebook \
         --source_path "$SRC_DIR/$i.md" \
         --dict_label "" \
-        --ebook_format mobi \
+        --output_format mobi \
         --output_path "$OUT_DIR/$i.mobi" \
         --mobi_compression 0 \
         --kindlegen_path "$KINDLEGEN_PATH"
 
     cargo run -- markdown_to_ebook \
         --source_path "$SRC_DIR/$i.md" \
         --dict_label "" \
-        --ebook_format epub \
+        --output_format epub \
         --output_path "$OUT_DIR/$i.epub"
 
+    # TODO epubcheck
+
+    stardict_out="$OUT_DIR/$i-stardict"
+    mkdir -p "$stardict_out"
+
+    cargo run -- markdown_to_stardict_xml \
+        --source_path "$SRC_DIR/$i.md" \
+        --output_path "$stardict_out/$i.xml"
+
+    cd "$stardict_out"
+    $STARDICT_TEXT2BIN "$i.xml" "$i.ifo"
+    rm "$i.xml"
+    cd ..
+    zip -r "$i-stardict.zip" "$i-stardict"
+    rm "$i-stardict" -r
+
 done
 
 # === Combined ===
 
+cd "$PROJ_ROOT"
+
+name="combined-dictionary"
+
 cargo run -- markdown_to_ebook \
-    --source_paths_list ./scripts/combined_dict_md_paths.txt \
     --title "Combined Pali - English Dictionary" \
-    --ebook_format mobi \
-    --output_path "$OUT_DIR/combined-dictionary.mobi" \
+    --source_paths_list ./scripts/combined_dict_md_paths.txt \
+    --output_format mobi \
+    --output_path "$OUT_DIR/$name.mobi" \
     --mobi_compression 0 \
     --kindlegen_path "$KINDLEGEN_PATH"
 
 cargo run -- markdown_to_ebook \
+    --title "Combined Pali - English Dictionary" \
     --source_paths_list ./scripts/combined_dict_md_paths.txt \
+    --output_format epub \
+    --output_path "$OUT_DIR/$name.epub"
+
+# TODO epubcheck
+
+stardict_out="$OUT_DIR/$name-stardict"
+mkdir -p "$stardict_out"
+
+cargo run -- markdown_to_stardict_xml \
     --title "Combined Pali - English Dictionary" \
-    --ebook_format epub \
-    --output_path "$OUT_DIR/combined-dictionary.epub"
+    --source_paths_list ./scripts/combined_dict_md_paths.txt \
+    --output_path "$stardict_out/$name.xml"
+
+cd "$stardict_out"
+$STARDICT_TEXT2BIN "$name.xml" "$name.ifo"
+rm "$name.xml"
+cd ..
+zip -r "$name-stardict.zip" "$name-stardict"
+rm "$name-stardict" -r
 
diff --git a/scripts/combined_to_mobi.sh b/scripts/combined_to_mobi.sh
@@ -3,7 +3,7 @@
 cargo run -- markdown_to_ebook \
 	--source_paths_list ./scripts/combined_dict_md_paths.txt \
 	--title "Combined Pali - English Dictionary" \
-	--ebook_format mobi \
+	--output_format mobi \
 	--output_path ../simsapa-dictionary_releases/combined-dictionary.mobi \
 	--mobi_compression 0 \
 	--dont_remove_generated_files \

diff --git a/scripts/json_to_md.sh b/scripts/json_to_md.sh
@@ -2,25 +2,67 @@
 
 #RUST_LOG=sources_to_markdown=info cargo run --bin sources_to_markdown 2>&1 | tee sources_to_markdown.log
 
-SC_ROOT="$HOME/src/suttacentral-2018-09-03/sc-data/dictionaries/en"
+SC_ROOT="$HOME/src/suttacentral-2019-11-15/sc-data/dictionaries/en"
 OUT_DIR="../simsapa-dictionary-data"
 
+# === Dhammika ===
+
+cargo run -- suttacentral_json_to_markdown \
+	--title "Dhammika Pali - English Dictionary" \
+	--dict_label Dhammika \
+	--json_path "$SC_ROOT/dhammika.json" \
+	--output_path "$OUT_DIR/dhammika.md"
+
 cargo run -- suttacentral_json_to_markdown \
+	--title "Dhammika Pali - English Dictionary" \
+	--dict_label Dhammika \
+	--dont_process \
 	--json_path "$SC_ROOT/dhammika.json" \
-	--output_path "$OUT_DIR/dhammika.md" \
-	--dict_label Dhammika
+	--output_path "$OUT_DIR/dhammika_unprocessed.md"
+
+# === DPPN ===
+
+cargo run -- suttacentral_json_to_markdown \
+	--title "Dictionary of Pali Proper Names (DPPN)" \
+	--dict_label DPPN \
+	--json_path "$SC_ROOT/dppn.json" \
+	--output_path "$OUT_DIR/dppn.md"
 
 cargo run -- suttacentral_json_to_markdown \
+	--title "Dictionary of Pali Proper Names (DPPN)" \
+	--dict_label DPPN \
+	--dont_process \
 	--json_path "$SC_ROOT/dppn.json" \
-	--output_path "$OUT_DIR/dppn.md" \
-	--dict_label DPPN
+	--output_path "$OUT_DIR/dppn_unprocessed.md"
+
+# === NCPED ===
+
+cargo run -- suttacentral_json_to_markdown \
+	--title "New Concise Pali - English Dictionary (NCPED)" \
+	--dict_label NCPED \
+	--json_path "$SC_ROOT/ncped.json" \
+	--output_path "$OUT_DIR/ncped.md"
 
 cargo run -- suttacentral_json_to_markdown \
+	--title "New Concise Pali - English Dictionary (NCPED)" \
+	--dict_label NCPED \
+	--dont_process \
 	--json_path "$SC_ROOT/ncped.json" \
-	--output_path "$OUT_DIR/ncped.md" \
-	--dict_label NCPED
+	--output_path "$OUT_DIR/ncped_unprocessed.md"
+
+# === PTS ===
+
+cargo run -- suttacentral_json_to_markdown \
+	--title "Pali Text Society Pali - English Dictionary (PTS)" \
+	--dict_label PTS \
+	--dont_remove_see_also \
+	--json_path "$SC_ROOT/pts.json" \
+	--output_path "$OUT_DIR/pts.md"
 
 cargo run -- suttacentral_json_to_markdown \
+	--title "Pali Text Society Pali - English Dictionary (PTS)" \
+	--dict_label PTS \
+	--dont_process \
+	--dont_remove_see_also \
 	--json_path "$SC_ROOT/pts.json" \
-	--output_path "$OUT_DIR/pts.md" \
-	--dict_label PTS
+	--output_path "$OUT_DIR/pts_unprocessed.md"
diff --git a/scripts/ncped_json_to_md.sh b/scripts/ncped_json_to_md.sh
@@ -0,0 +1,12 @@
+#!/bin/bash
+
+#RUST_LOG=sources_to_markdown=info cargo run --bin sources_to_markdown 2>&1 | tee sources_to_markdown.log
+
+SC_ROOT="$HOME/src/suttacentral-2019-11-15/sc-data/dictionaries/en"
+OUT_DIR="../simsapa-dictionary-data"
+
+cargo run -- suttacentral_json_to_markdown \
+	--json_path "$SC_ROOT/ncped.json" \
+	--output_path "$OUT_DIR/ncped.md" \
+	--dict_label NCPED
+