feat: Added pytesseract wrapper

pyproject.toml

@@ -60,6 +60,7 @@
     "networkx",
     "pydantic",
     "psutil",
+    "pytesseract",
   ]
 
   [project.optional-dependencies]

requirements/dev_requirements.txt

@@ -73,6 +73,8 @@ docutils==0.17.1
     #   sphinx
 entrypoints==0.4
     # via jupyter-client
+exceptiongroup==1.1.3
+    # via pytest
 executing==1.2.0
     # via stack-data
 fastjsonschema==2.16.2
@@ -97,6 +99,7 @@ importlib-metadata==5.1.0
     # via
     #   jupyter-cache
     #   myst-nb
+    #   sphinx
 iniconfig==1.1.1
     # via pytest
 ipykernel==6.19.1
@@ -222,6 +225,7 @@ packaging==22.0
     #   matplotlib
     #   pikepdf
     #   pydata-sphinx-theme
+    #   pytesseract
     #   pytest
     #   setuptools-scm
     #   spacy
@@ -252,6 +256,7 @@ pillow==9.3.0
     # via
     #   matplotlib
     #   pikepdf
+    #   pytesseract
     #   sec-certs (./../pyproject.toml)
 pip-tools==6.11.0
     # via sec-certs (./../pyproject.toml)
@@ -304,6 +309,8 @@ pyrsistent==0.19.2
     # via jsonschema
 pysankeybeta==1.4.0
     # via sec-certs (./../pyproject.toml)
+pytesseract==0.3.10
+    # via sec-certs (./../pyproject.toml)
 pytest==7.2.0
     # via
     #   pytest-cov
@@ -434,7 +441,9 @@ tomli==2.0.1
     #   pytest
     #   setuptools-scm
 tornado==6.3.3
-    # via setuptools-scm
+    # via
+    #   ipykernel
+    #   jupyter-client
 tqdm==4.64.1
     # via
     #   sec-certs (./../pyproject.toml)
@@ -464,10 +473,12 @@ types-urllib3==1.26.25.4
     # via types-requests
 typing-extensions==4.4.0
     # via
+    #   black
     #   mypy
     #   myst-nb
     #   myst-parser
     #   pydantic
+    #   pypdf
     #   setuptools-scm
 urllib3==1.26.13
     # via requests

requirements/requirements.txt

@@ -128,6 +128,7 @@ packaging==22.0
     #   ipykernel
     #   matplotlib
     #   pikepdf
+    #   pytesseract
     #   setuptools-scm
     #   spacy
 pandas==1.5.2
@@ -152,6 +153,7 @@ pillow==9.3.0
     # via
     #   matplotlib
     #   pikepdf
+    #   pytesseract
     #   sec-certs (./../pyproject.toml)
 pkgconfig==1.5.5
     # via sec-certs (./../pyproject.toml)
@@ -189,6 +191,8 @@ pyrsistent==0.19.2
     # via jsonschema
 pysankeybeta==1.4.0
     # via sec-certs (./../pyproject.toml)
+pytesseract==0.3.10
+    # via sec-certs (./../pyproject.toml)
 python-dateutil==2.8.2
     # via
     #   jupyter-client
@@ -275,6 +279,7 @@ typer==0.7.0
 typing-extensions==4.4.0
     # via
     #   pydantic
+    #   pypdf
     #   setuptools-scm
 urllib3==1.26.13
     # via requests

requirements/test_requirements.txt

@@ -52,6 +52,8 @@ distro==1.8.0
     # via tabula-py
 entrypoints==0.4
     # via jupyter-client
+exceptiongroup==1.1.3
+    # via pytest
 executing==1.2.0
     # via stack-data
 fonttools==4.38.0
@@ -136,6 +138,7 @@ packaging==22.0
     #   ipykernel
     #   matplotlib
     #   pikepdf
+    #   pytesseract
     #   pytest
     #   setuptools-scm
     #   spacy
@@ -161,6 +164,7 @@ pillow==9.3.0
     # via
     #   matplotlib
     #   pikepdf
+    #   pytesseract
     #   sec-certs (./../pyproject.toml)
 pkgconfig==1.5.5
     # via sec-certs (./../pyproject.toml)
@@ -200,6 +204,8 @@ pyrsistent==0.19.2
     # via jsonschema
 pysankeybeta==1.4.0
     # via sec-certs (./../pyproject.toml)
+pytesseract==0.3.10
+    # via sec-certs (./../pyproject.toml)
 pytest==7.2.0
     # via
     #   pytest-cov
@@ -295,6 +301,7 @@ typer==0.7.0
 typing-extensions==4.4.0
     # via
     #   pydantic
+    #   pypdf
     #   setuptools-scm
 urllib3==1.26.13
     # via requests

src/sec_certs/utils/pdf.py

@@ -11,6 +11,7 @@
 
 import pdftotext
 import pikepdf
+import pytesseract
 
 from sec_certs import constants
 from sec_certs.constants import (
@@ -51,13 +52,16 @@ def ocr_pdf_file(pdf_path: Path) -> str:
         )
         if ppm.returncode != 0:
             raise ValueError(f"pdftoppm failed: {ppm.returncode}")
+
         for ppm_path in map(Path, glob.glob(str(tmppath / "image*.ppm"))):
             base = ppm_path.with_suffix("")
-            tes = subprocess.run(
-                ["tesseract", "-l", "eng+deu+fra", ppm_path, base], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL
-            )
-            if tes.returncode != 0:
-                raise ValueError(f"tesseract failed: {tes.returncode}")
+            content = pytesseract.image_to_string(ppm_path, lang="eng+deu+fra")
+
+            if content:
+                with Path(base.with_suffix(".txt")).open("w") as file:
+                    file.write(content)
+            else:
+                raise ValueError(f"OCR failed for document {ppm_path}. Check document manually")
 
         contents = ""