From 1a2dc2565b504055a33f1af0d15a333025bb5c94 Mon Sep 17 00:00:00 2001
From: scossu <stefano@cossu.cc>
Date: Wed, 28 Feb 2024 08:38:25 -0500
Subject: [PATCH] Yiddish transliteration via submodules.

---
 .github/workflows/push-test-image.yml    |  2 +-
 .gitmodules                              |  4 ++
 Dockerfile                               | 17 ++++----
 ext/yiddish                              |  1 +
 requirements.txt                         |  1 +
 scriptshifter/hooks/yiddish_/__init__.py | 51 ++++++++++++++++++++++++
 scriptshifter/tables/data/index.yml      |  2 +
 scriptshifter/tables/data/yiddish.yml    | 21 ++++++++++
 8 files changed, 90 insertions(+), 9 deletions(-)
 create mode 160000 ext/yiddish
 create mode 100644 scriptshifter/hooks/yiddish_/__init__.py
 create mode 100644 scriptshifter/tables/data/yiddish.yml

diff --git a/.github/workflows/push-test-image.yml b/.github/workflows/push-test-image.yml
index 2bcc042..786b893 100644
--- a/.github/workflows/push-test-image.yml
+++ b/.github/workflows/push-test-image.yml
@@ -2,7 +2,7 @@ name: Push test image to Docker Hub.
 on:
   push:
     branch:
-      - "main"
+      - "test"
 
 env:
   DOCKER_USER: lcnetdev
diff --git a/.gitmodules b/.gitmodules
index 45cbf19..cc82df3 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -1,3 +1,7 @@
 [submodule "ext/arabic_rom"]
 	path = ext/arabic_rom
 	url = https://github.com/fadhleryani/Arabic_ALA-LC_Romanization.git
+[submodule "ext/yiddish"]
+	path = ext/yiddish
+	url = https://github.com/ibleaman/yiddish.git
+	branch = loc
diff --git a/Dockerfile b/Dockerfile
index 1ff0a43..578751e 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -6,14 +6,6 @@ RUN apt install -y build-essential tzdata gfortran libopenblas-dev libboost-all-
 ENV TZ=America/New_York
 ENV _workroot "/usr/local/scriptshifter/src"
 
-WORKDIR ${_workroot}
-COPY requirements.txt ./
-RUN pip install --no-cache-dir -r requirements.txt
-
-# Remove development packages.
-RUN apt remove -y build-essential
-RUN apt autoremove -y
-
 RUN addgroup --system www
 RUN adduser --system www
 RUN gpasswd -a www www
@@ -22,6 +14,15 @@ COPY entrypoint.sh uwsgi.ini wsgi.py ./
 COPY ext ./ext/
 COPY scriptshifter ./scriptshifter/
 
+WORKDIR ${_workroot}
+COPY requirements.txt ./
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Remove development packages.
+RUN apt remove -y build-essential git
+RUN apt autoremove -y
+RUN rm -rf ext/yiddish
+
 RUN chmod +x ./entrypoint.sh
 RUN chown -R www:www ${_workroot} .
 
diff --git a/ext/yiddish b/ext/yiddish
new file mode 160000
index 0000000..9bf22c5
--- /dev/null
+++ b/ext/yiddish
@@ -0,0 +1 @@
+Subproject commit 9bf22c55ca76710940e141de5d88922a9f55ed1f
diff --git a/requirements.txt b/requirements.txt
index 3136aa2..ccd5d4c 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -7,3 +7,4 @@ python-dotenv>=1.0,<2
 pyyaml>=6.0,<7
 repackage>=0.7.3
 uwsgi>=2.0,<2.1
+./ext/yiddish
diff --git a/scriptshifter/hooks/yiddish_/__init__.py b/scriptshifter/hooks/yiddish_/__init__.py
new file mode 100644
index 0000000..22ca3a8
--- /dev/null
+++ b/scriptshifter/hooks/yiddish_/__init__.py
@@ -0,0 +1,51 @@
+# @package ext
+
+__doc__ = """
+Yiddish transliteration module.
+
+Courtesy of Isaac Bleaman and Asher Lewis.
+
+https://github.com/ibleaman/yiddish.git
+
+Note the underscore in the module name to disambiguate with the `yiddish`
+external package name.
+"""
+
+
+from yiddish import detransliterate, transliterate
+
+from scriptshifter.exceptions import BREAK
+from scriptshifter.tools import capitalize
+
+
+def s2r_post_config(ctx):
+    """
+    Script to Roman.
+    """
+
+    rom = transliterate(
+            ctx.src, loc=True,
+            loshn_koydesh=ctx.options.get("loshn_koydesh"))
+
+    if ctx.options["capitalize"] == "all":
+        rom = capitalize(rom)
+    elif ctx.options["capitalize"] == "first":
+        rom = rom[0].upper() + rom[1:]
+
+    ctx.dest = rom
+
+    return BREAK
+
+
+def r2s_post_config(ctx):
+    """
+    Roman to script.
+
+    NOTE: This doesn't support the `loc` option.
+    """
+
+    ctx.dest = detransliterate(
+            ctx.src,
+            loshn_koydesh=ctx.options.get("loshn_koydesh"))
+
+    return BREAK
diff --git a/scriptshifter/tables/data/index.yml b/scriptshifter/tables/data/index.yml
index f6d89ac..889ba8f 100644
--- a/scriptshifter/tables/data/index.yml
+++ b/scriptshifter/tables/data/index.yml
@@ -154,5 +154,7 @@ uzbek_cyrillic:
   name: Uzbek (Cyrillic)
 yakut_cyrillic:
   name: Yakut (Cyrillic)
+yiddish:
+  name: Yiddish
 yuit_cyrillic:
   name: Yuit (Cyrillic)
diff --git a/scriptshifter/tables/data/yiddish.yml b/scriptshifter/tables/data/yiddish.yml
new file mode 100644
index 0000000..c55c431
--- /dev/null
+++ b/scriptshifter/tables/data/yiddish.yml
@@ -0,0 +1,21 @@
+general:
+  name: Yiddish
+
+options:
+  - id: loshn_koydesh
+    label: Loshn Koydesh
+    description: [TODO]
+    type: boolean
+    default: false
+
+script_to_roman:
+  hooks:
+    post_config:
+      -
+        - yiddish_.s2r_post_config
+
+roman_to_script:
+  hooks:
+    post_config:
+      -
+        - yiddish_.r2s_post_config