Merge pull request #104 from googlefonts/test_strings

Add test strings (attempt 2)

.github/workflows/test.yml

@@ -28,5 +28,4 @@ jobs:
           pip install pytest
       - name: Run Tests
         run: |
-          pytest ./tests/testglyphdata.py
-          pytest ./tests/testusage.py
+          pytest ./tests

Lib/glyphsets/__init__.py

@@ -10,6 +10,7 @@
 import logging
 import unicodedata2 as uni
 from glyphsLib.glyphdata import get_glyph
+from copy import deepcopy as copy
 
 try:
     from ._version import version as __version__  # type: ignore
@@ -18,12 +19,13 @@
 
 
 DATA_FP = os.path.join(os.path.dirname(__file__), "data.json")
+TEST_STRINGS_DATA = os.path.join(os.path.dirname(__file__), "test_strings.json")
 log = logging.getLogger(__file__)
 
 
 class _GFGlyphData:
-    def __init__(self, data=json.load(open(DATA_FP))):
-        self._data = data
+    def __init__(self, data=json.load(open(DATA_FP, encoding="utf8"))):
+        self._data = copy(data)
         self._in_use = set(g["nice_name"] for g in self._data["glyphs"])
 
     def save(self, fp=DATA_FP):
@@ -39,7 +41,7 @@ def from_json(cls, fp):
     def __getitem__(self, k):
         return self._data[k]
 
-    def missing_glyphsets_in_font(self, ttFont, threshold=0.8):
+    def glyphsets_in_font(self, ttFont):
         glyphs_in_font = set(ttFont.getGlyphOrder())
         unicodes_in_font = set(ttFont.getBestCmap().keys())
         res = {}
@@ -57,7 +59,17 @@ def missing_glyphsets_in_font(self, ttFont, threshold=0.8):
                     res[glyphset]["has"].append(g)
                 else:
                     res[glyphset]["missing"].append(g)
+        return res
+
+    def glyphsets_fulfilled(self, ttFont):
+        res = self.glyphsets_in_font(ttFont)
+        return {
+            k: len(v["has"]) / (len(v["has"]) + len(v["missing"]))
+            for k, v in res.items()
+        }
 
+    def missing_glyphsets_in_font(self, ttFont, threshold=0.8):
+        res = self.glyphsets_in_font(ttFont)
         fulfilled = {
             k: len(v["has"]) / (len(v["has"]) + len(v["missing"]))
             for k, v in res.items()
@@ -233,3 +245,28 @@ def build_fea(self, glyphset):
 
 
 GFGlyphData = _GFGlyphData()
+
+
+class _TestDocData:
+    def __init__(
+        self,
+        data=json.load(open(TEST_STRINGS_DATA, encoding="utf8")),
+        glyphsets=GFGlyphData,
+    ):
+        self._data = data
+        self._glyphsets = glyphsets
+
+    def test_strings_in_font(self, ttFont, threshold=0.95):
+        res = {}
+        glyphsets_in_font = self._glyphsets.glyphsets_fulfilled(ttFont)
+        for glyphset, coverage in glyphsets_in_font.items():
+            if coverage < threshold:
+                continue
+            test_strings = self._data.get(glyphset)
+            if not test_strings:
+                continue
+            res[glyphset] = test_strings
+        return res
+
+
+GFTestData = _TestDocData()

Lib/glyphsets/test_strings.json

@@ -0,0 +1,35 @@
+{
+    "GF_Latin_Kernel": {
+        "Letter Groups": [
+            "EFHIJLTU", 
+            "AVWXZ",
+            "KMNY",
+            "COQS",
+            "BPRDG",
+            "fijltr",
+            "kvwxyz",
+            "hmnu",
+            "ceogs",
+            "abdpq"
+        ],
+        "Spacing": [
+            "HHOHHOOHOO HIHOIO",
+            "uuonouonoo ninoioolonlniuo",
+            "010 080 030 020 050 060 070"
+        ],
+        "Kerning": [
+            "HHAVAHH WAFAPATAUAYALYTJĿLVÆ HH",
+            "TiTíTîTïTìTĭTīTĩT Vă Tä Tü Tõ Tŭ Tř",
+            "HHTan oTo yTy vAv oVo eYe Ly Ky Ac",
+            "nin non nvn ovo oxo ayn fl fi fj ft gj ko rento",
+            "ďá ďu ŀl gj ľk włw yły ółr fħ ílïlîl",
+            "H.Y-Y.T,F.P,V:Y„V-T-A*A’A“A’A-AL–L—L-",
+            "n.r.r,y.y,(o)(i)(d)(j)(f)[j]{f}",
+            "080 076 474 973 94 7078",
+            "P4T47A .47,9.7-»4",
+            "42°21′29″N 71°03′49″W",
+            "d² m³ 15°C/26°F",
+            "0/0\\0"
+        ]
+    }
+}

setup.py

@@ -42,7 +42,7 @@
     package_data={'glyphsets': [
                       "encodings/*.nam",
                       "encodings/GF Glyph Sets/*.nam",
-                      "data.json",
+                      "*.json",
                   ]
                  },
     entry_points={"console_scripts": ["glyphsets = glyphsets.__main__:main"]},

tests/__init__.py

@@ -0,0 +1,64 @@
+from glyphsets import _GFGlyphData, _TestDocData
+from fontTools.ttLib import TTFont
+import os
+import pytest
+
+DATA_FP = os.path.join(os.path.dirname(__file__), "data")
+
+glyphs_data = {
+    "glyphs": [
+        {
+            "nice_name": "A",
+            "production_name": "A",
+            "character": "A",
+            "unicode": 65,
+            "glyphsets": [
+                "GFLatinIPA",
+                "GFLatinAfrican",
+                "GFLatinCore",
+                "GFLatinKernel",
+                "GFLatinMinorities",
+                "GFLatinPlus",
+                "GFLatinVietnamese",
+            ],
+        },
+        {
+            "nice_name": "brevecomb_acutecomb",
+            "production_name": "uni03060301",
+            "character": None,
+            "unicode": None,
+            "glyphsets": ["GFLatinVietnamese"],
+        },
+        {
+            "nice_name": "fdotaccent",
+            "production_name": "uni1E1F",
+            "character": "\u1e1f",
+            "unicode": 7711,
+            "glyphsets": ["GFLatinAfrican"],
+        },
+    ]
+}
+
+
+@pytest.fixture
+def glyph_data():
+    return _GFGlyphData(glyphs_data)
+
+
+@pytest.fixture
+def test_string_data():
+    data = {
+        "GFLatinKernel": {
+            "Spacing": ["AAA"],
+        },
+        "GFLatinAfrican": {
+            "Spacing": ["BBB"],
+        },
+    }
+    return _TestDocData(data, _GFGlyphData(glyphs_data))
+
+
+@pytest.fixture
+def ttFont():
+    fp = os.path.join(DATA_FP, "MavenPro[wght].ttf")
+    return TTFont(fp)

tests/testglyphdata.py → tests/test_glyphdata.py

@@ -1,55 +1,7 @@
 import pytest
-from fontTools.ttLib import TTFont
-from glyphsets import _GFGlyphData
 from glyphsLib import GSFont, GSGlyph
 from defcon import Font
-import os
-
-DATA_FP = os.path.join(os.path.dirname(__file__), "data")
-
-
-@pytest.fixture
-def db():
-    data = {
-        "glyphs": [
-            {
-                "nice_name": "A",
-                "production_name": "A",
-                "character": "A",
-                "unicode": 65,
-                "glyphsets": [
-                    "GFLatinIPA",
-                    "GFLatinAfrican",
-                    "GFLatinCore",
-                    "GFLatinKernel",
-                    "GFLatinMinorities",
-                    "GFLatinPlus",
-                    "GFLatinVietnamese",
-                ],
-            },
-            {
-                "nice_name": "brevecomb_acutecomb",
-                "production_name": "uni03060301",
-                "character": None,
-                "unicode": None,
-                "glyphsets": ["GFLatinVietnamese"],
-            },
-            {
-                "nice_name": "fdotaccent",
-                "production_name": "uni1E1F",
-                "character": "\u1e1f",
-                "unicode": 7711,
-                "glyphsets": ["GFLatinAfrican"],
-            },
-        ]
-    }
-    return _GFGlyphData(data)
-
-
-@pytest.fixture
-def ttFont():
-    fp = os.path.join(DATA_FP, "MavenPro[wght].ttf")
-    return TTFont(fp)
+from . import *
 
 
 def glyphs_src1():
@@ -85,8 +37,8 @@ def ufo_src1():
         (["foobar"], []),
     ],
 )
-def test_build_name_file(db, test_input, expected):
-    nam_data = db.build_nam_file(test_input)
+def test_build_name_file(glyph_data, test_input, expected):
+    nam_data = glyph_data.build_nam_file(test_input)
     assert nam_data == expected
 
 
@@ -101,8 +53,8 @@ def test_build_name_file(db, test_input, expected):
         (["foobar"], []),
     ],
 )
-def test_glyphsapp_filter_lists(db, test_input, expected):
-    filter_list = db.build_glyphsapp_filter_list(test_input)
+def test_glyphsapp_filter_lists(glyph_data, test_input, expected):
+    filter_list = glyph_data.build_glyphsapp_filter_list(test_input)
     assert filter_list == expected
 
 
@@ -130,8 +82,8 @@ def test_glyphsapp_filter_lists(db, test_input, expected):
         (ufo_src1(), ["GFLatinCore"], ["A"]),
     ],
 )
-def test_update_source_glyphset(db, src, test_input, expected):
-    db.update_source_glyphset(src, test_input)
+def test_update_source_glyphset(glyph_data, src, test_input, expected):
+    glyph_data.update_source_glyphset(src, test_input)
 
     if isinstance(src, GSFont):
         assert [g.name for g in src.glyphs] == expected
@@ -141,41 +93,41 @@ def test_update_source_glyphset(db, src, test_input, expected):
         assert set(g.name for g in src) == set(expected)
 
 
-def test_update_db(db):
-    glyphs = [g["nice_name"] for g in db._data["glyphs"]]
+def test_update_glyph_data(glyph_data):
+    glyphs = [g["nice_name"] for g in glyph_data._data["glyphs"]]
     assert glyphs == ["A", "brevecomb_acutecomb", "fdotaccent"]
 
     # Add a new glyphset to an already existing glyph
     src1 = GSFont()
     src1.filepath = "NewGlyphSet.glyphs"
     src1.glyphs.append(GSGlyph("A"))
-    db.update_db_from_sources([src1])
-    assert "NewGlyphSet" in db["glyphs"][0]["glyphsets"]
+    glyph_data.update_db_from_sources([src1])
+    assert "NewGlyphSet" in glyph_data["glyphs"][0]["glyphsets"]
 
     # Add duplicate glyphs
-    db_size = len(db["glyphs"])
+    glyph_data_size = len(glyph_data["glyphs"])
     src2 = GSFont()
     src2.filepath = "GFLatinCore.glyphs"
     src2.glyphs.append(GSGlyph("A"))
     src2.glyphs.append(GSGlyph("brevecomb_acutecomb"))
     src2.glyphs.append(GSGlyph("fdotaccent"))
-    db.update_db_from_sources([src2])
-    assert len(db["glyphs"]) == db_size
+    glyph_data.update_db_from_sources([src2])
+    assert len(glyph_data["glyphs"]) == glyph_data_size
 
     # Add new glyphs
     src3 = GSFont()
     src3.filepath = "Greek.glyphs"
     src3.glyphs.append(GSGlyph("Alpha"))
     src3.glyphs.append(GSGlyph("Beta"))
-    db.update_db_from_sources([src3])
-    assert len(db["glyphs"]) == db_size + 2
+    glyph_data.update_db_from_sources([src3])
+    assert len(glyph_data["glyphs"]) == glyph_data_size + 2
 
     # TODO ufo sources
 
 
-def test_glyphsets_missing_in_font(db, ttFont):
+def test_glyphsets_missing_in_font(glyph_data, ttFont):
     # TODO use fonttools fontbuilder and make a font from scratch
-    missing = db.missing_glyphsets_in_font(ttFont, threshold=0)
+    missing = glyph_data.missing_glyphsets_in_font(ttFont, threshold=0)
     assert missing == {
         "GFLatinAfrican": [
             {

tests/test_teststrings.py

@@ -0,0 +1,11 @@
+import pytest
+from . import *
+
+
+def test_test_strings_in_font(test_string_data, ttFont):
+    res = test_string_data.test_strings_in_font(ttFont)
+    assert "GFLatinKernel" in res
+
+    # Drop threshold so we pick up GFLatinAfrican
+    res = test_string_data.test_strings_in_font(ttFont, 0.4)
+    assert "GFLatinAfrican" in res