Merge pull request #6 from fedecosta/develop

added ce and ba files

data/ca-ba/language.yml

@@ -0,0 +1,50 @@
+---
+
+language:
+  name: "Balear Catalan"
+  code: "ca-ba"
+  phonemes: !env "${config_dir}/phonemes.txt"
+  keep_stress: true
+
+lexicon: !env "${config_dir}/lexicon.db"
+
+g2p:
+  model: !env "${config_dir}/g2p.fst"
+
+symbols:
+  casing: "lower"
+  number_regex: "^-?\\d+([,.]\\d+)*$"
+  token_split: "\\s+"
+  token_join: " "
+  minor_breaks:
+    - ","
+    - ":"
+    - ";"
+    - "..."
+  major_breaks:
+    - "."
+    - "?"
+    - "!"
+  replace:
+    "[\\<\\>\\(\\)\\[\\]\"]+": ""
+    "\\B'": "\""
+    "'\\B": "\""
+    "’": "'"
+    "'": ""
+    "-": ""
+    "l·l": "l"
+  punctuations:
+    - "\""
+    - "„"
+    - "“"
+    - "”"
+    - "«"
+    - "»"
+    - ","
+    - ":"
+    - ";"
+    - "."
+    - "?"
+    - "¿" 
+    - "!"
+    - "¡"

data/ca-ba/phonemes.txt

@@ -0,0 +1,44 @@
+# https://en.wikipedia.org/wiki/Catalan_phonology
+# Catalan Central accent phonemes
+
+p [p]ala
+b [b]ala
+t [t]ela
+d [d]onar
+k [k]ala
+ɡ [g]ala
+m [m]ala
+ɲ fa[ng]
+β aca[b]a
+ð ca[d]a
+ɣ ama[g]ar
+f [f]als
+v a[f]ganès
+s [s]ala
+z ca[s]a
+ʃ [x]oc
+ʒ mà[g]ic
+tʃ co[tx]e
+dʒ me[tg]e
+l [l]íquid
+ʎ [ll]amp
+r ca[rr]o
+ɾ ca[r]a
+w ve[u]en
+uw ca[u]re
+j ca[i]re
+y [i]a[i]a
+n [n]ena
+ŋ pi[n]güí
+ts po[ts]er
+dz do[tz]e
+
+# Vowels
+i r[i]c
+e c[e]c
+ɛ s[e]c
+a s[a]c
+ɔ f[o]c
+o s[ó]c
+u s[u]c
+ə [a]mor

data/ca-ce/language.yml

@@ -0,0 +1,50 @@
+---
+
+language:
+  name: "Central Catalan"
+  code: "ca-ce"
+  phonemes: !env "${config_dir}/phonemes.txt"
+  keep_stress: true
+
+lexicon: !env "${config_dir}/lexicon.db"
+
+g2p:
+  model: !env "${config_dir}/g2p.fst"
+
+symbols:
+  casing: "lower"
+  number_regex: "^-?\\d+([,.]\\d+)*$"
+  token_split: "\\s+"
+  token_join: " "
+  minor_breaks:
+    - ","
+    - ":"
+    - ";"
+    - "..."
+  major_breaks:
+    - "."
+    - "?"
+    - "!"
+  replace:
+    "[\\<\\>\\(\\)\\[\\]\"]+": ""
+    "\\B'": "\""
+    "'\\B": "\""
+    "’": "'"
+    "'": ""
+    "-": ""
+    "l·l": "l"
+  punctuations:
+    - "\""
+    - "„"
+    - "“"
+    - "”"
+    - "«"
+    - "»"
+    - ","
+    - ":"
+    - ";"
+    - "."
+    - "?"
+    - "¿" 
+    - "!"
+    - "¡"

data/ca-ce/phonemes.txt

@@ -0,0 +1,44 @@
+# https://en.wikipedia.org/wiki/Catalan_phonology
+# Catalan Central accent phonemes
+
+p [p]ala
+b [b]ala
+t [t]ela
+d [d]onar
+k [k]ala
+ɡ [g]ala
+m [m]ala
+ɲ fa[ng]
+β aca[b]a
+ð ca[d]a
+ɣ ama[g]ar
+f [f]als
+v a[f]ganès
+s [s]ala
+z ca[s]a
+ʃ [x]oc
+ʒ mà[g]ic
+tʃ co[tx]e
+dʒ me[tg]e
+l [l]íquid
+ʎ [ll]amp
+r ca[rr]o
+ɾ ca[r]a
+w ve[u]en
+uw ca[u]re
+j ca[i]re
+y [i]a[i]a
+n [n]ena
+ŋ pi[n]güí
+ts po[ts]er
+dz do[tz]e
+
+# Vowels
+i r[i]c
+e c[e]c
+ɛ s[e]c
+a s[a]c
+ɔ f[o]c
+o s[ó]c
+u s[u]c
+ə [a]mor

gruut-lang-ca/LANGUAGE

@@ -0,0 +1 @@
+ca-ce Catalan

gruut-lang-ca/README.md

@@ -0,0 +1,3 @@
+# gruut Catalan
+
+Language-specific files for Catalan (ca) in [gruut](https://github.com/rhasspy/gruut)

gruut-lang-ca/gruut_lang_ca/VERSION

@@ -0,0 +1 @@
+0.0.0

gruut-lang-ca/gruut_lang_ca/__init__.py

@@ -0,0 +1,22 @@
+"""Catalan language resources"""
+import os
+import typing
+from pathlib import Path
+
+try:
+    import importlib.resources
+
+    files = importlib.resources.files
+except (ImportError, AttributeError):
+    # Backport for Python < 3.9
+    import importlib_resources  # type: ignore
+
+    files = importlib_resources.files
+
+_PACKAGE = "gruut_lang_ca"
+_DIR = Path(typing.cast(os.PathLike, files(_PACKAGE)))
+
+
+def get_lang_dir() -> Path:
+    """Get directory with language resources"""
+    return _DIR

gruut-lang-ca/gruut_lang_ca/phonemes.txt

@@ -0,0 +1,44 @@
+# https://en.wikipedia.org/wiki/Catalan_phonology
+# Catalan Central accent phonemes
+
+p [p]ala
+b [b]ala
+t [t]ela
+d [d]onar
+k [k]ala
+ɡ [g]ala
+m [m]ala
+ɲ fa[ng]
+β aca[b]a
+ð ca[d]a
+ɣ ama[g]ar
+f [f]als
+v a[f]ganès
+s [s]ala
+z ca[s]a
+ʃ [x]oc
+ʒ mà[g]ic
+tʃ co[tx]e
+dʒ me[tg]e
+l [l]íquid
+ʎ [ll]amp
+r ca[rr]o
+ɾ ca[r]a
+w ve[u]en
+uw ca[u]re
+j ca[i]re
+y [i]a[i]a
+n [n]ena
+ŋ pi[n]güí
+ts po[ts]er
+dz do[tz]e
+
+# Vowels
+i r[i]c
+e c[e]c
+ɛ s[e]c
+a s[a]c
+ɔ f[o]c
+o s[ó]c
+u s[u]c
+ə [a]mor

gruut-lang-ca/setup.py

@@ -0,0 +1,58 @@
+"""Setup file for gruut_lang_ca"""
+from pathlib import Path
+
+import setuptools
+
+module_name = "gruut_lang_ca"
+
+this_dir = Path(__file__).parent
+module_dir = this_dir / module_name
+
+# -----------------------------------------------------------------------------
+
+# Load README in as long description
+long_description: str = ""
+readme_path = this_dir / "README.md"
+if readme_path.is_file():
+    long_description = readme_path.read_text(encoding="utf-8")
+
+version_path = module_dir / "VERSION"
+with open(version_path, "r", encoding="utf-8") as version_file:
+    version = version_file.read().strip()
+
+
+# Extra package data files
+extra_files = []
+maybe_extra_files = ["pos/model.crf", "pos/postagger.model"]
+for maybe_extra_str in maybe_extra_files:
+    extra_path = module_dir / maybe_extra_str
+    if extra_path.is_file():
+        extra_files.append(maybe_extra_str)
+
+# -----------------------------------------------------------------------------
+
+setuptools.setup(
+    name=module_name,
+    description="Catalan language files for gruut tokenizer/phonemizer",
+    version=version,
+    author="Michael Hansen",
+    author_email="[email protected]",
+    url="https://github.com/rhasspy/gruut",
+    packages=setuptools.find_packages(),
+    package_data={
+        module_name: [
+            "VERSION",
+            "lexicon.db",
+            "g2p/model.crf",
+            #"espeak/lexicon.db",
+            #"espeak/g2p/model.crf",
+        ]
+        + extra_files
+    },
+    classifiers=[
+        "Programming Language :: Python :: 3",
+        "License :: OSI Approved :: MIT License",
+    ],
+    long_description=long_description,
+    long_description_content_type="text/markdown",
+)