Tip: `tree_sitter_typescript` on `aarch64 Android` - how to compile for docling

[Manamama]

Summary:

docling (more precisely: docling-core[chunking] → semchunk → tree-sitter-typescript), at least as of today, pulls an old / incompatible version that failed to build on Termux/Android.

Breakdown of responsibility

Component	Version pulled / attempted	Blame level	Reason
docling	2.77.0 (latest at time)	Medium	Depends on docling-core[chunking] which requires semchunk
docling-core[chunking]	<3.0.0, >=2.66.0	High	Explicitly requires semchunk<3.0.0,>=2.2.0 for code chunking
semchunk	2.2.2	High	Depends on tree-sitter-typescript (no version pin → pip picks latest)
tree-sitter-typescript	0.23.2 (git HEAD)	Very High	Build fails on Termux because missing tree_sitter/parser.h headers or clang/Android quirks in generated parser.c (TSFieldMapSlice unknown type)
tree-sitter core	Not installed as dev pkg	Medium	Termux has runtime tree-sitter but no -dev package → pip can't find headers for source build
pip / build env	Modern (26.0.1)	Low	Tries source build when no wheel exists for android_24_arm64_v8a

Root cause chain:

• docling → wants chunking → pulls semchunk 2.2.2
• semchunk → wants tree-sitter-typescript
• pip sees no pre-built wheel for tree-sitter-typescript on Android aarch64 → falls back to source build from git
• Source build → clang compile of tsx/src/parser.c → fails because tree-sitter headers are missing or incompatible with Termux's clang/Android sysroot

What to do

git clone https://github.com/tree-sitter/tree-sitter-typescript.git and pip install . this version. I guess pip install git+https://github.com/tree-sitter/tree-sitter-typescript.git@0.20.2 --no-build-isolation --no-deps would also do. May need tinkering with compile flags.

Symptoms:

If just pip install -U docling:

    dist._finalize_license_expression()
  /data/data/com.termux/files/usr/lib/python3.13/site-packages/setuptools/dist.py:765: SetuptoolsDeprecationWarning: License classifiers are deprecated.
  !!

          ********************************************************************************
          Please consider removing the following classifiers in favor of a SPDX license expression:

          License :: OSI Approved :: MIT License

          See https://packaging.python.org/en/latest/guides/writing-pyproject-toml/#license for details.
          ********************************************************************************

  !!
    self._finalize_license_expression()
  running bdist_wheel
  running build
  running build_py
  creating build/lib.android-24-arm64_v8a-cpython-313/tree_sitter_typescript
  copying bindings/python/tree_sitter_typescript/__init__.py -> build/lib.android-24-arm64_v8a-cpython-313/tree_sitter_typescript
  running egg_info
  writing bindings/python/tree_sitter_typescript.egg-info/PKG-INFO
  writing dependency_links to bindings/python/tree_sitter_typescript.egg-info/dependency_links.txt
  writing requirements to bindings/python/tree_sitter_typescript.egg-info/requires.txt
  writing top-level names to bindings/python/tree_sitter_typescript.egg-info/top_level.txt
  [03/09/26 09:57:41] ERROR    listing git files failed - pretending     git.py:26
                               there aren't any
  reading manifest file 'bindings/python/tree_sitter_typescript.egg-info/SOURCES.txt'
  adding license file 'LICENSE'
  writing manifest file 'bindings/python/tree_sitter_typescript.egg-info/SOURCES.txt'
  copying bindings/python/tree_sitter_typescript/__init__.pyi -> build/lib.android-24-arm64_v8a-cpython-313/tree_sitter_typescript
  copying bindings/python/tree_sitter_typescript/binding.c -> build/lib.android-24-arm64_v8a-cpython-313/tree_sitter_typescript
  copying bindings/python/tree_sitter_typescript/py.typed -> build/lib.android-24-arm64_v8a-cpython-313/tree_sitter_typescript
  running build_ext
  building '_binding' extension
  creating build/temp.android-24-arm64_v8a-cpython-313/bindings/python/tree_sitter_typescript
  creating build/temp.android-24-arm64_v8a-cpython-313/tsx/src
  creating build/temp.android-24-arm64_v8a-cpython-313/typescript/src
  clang -Wno-error=implicit-function-declaration -Wno-c2y-extensions -Wno-unused-variable -Wno-error=c2y-extensions -fvisibility=default -fPIC -DPy_LIMITED_API=0x03090000 -DPY_SSIZE_T_CLEAN -DTREE_SITTER_HIDE_SYMBOLS -Itypescript/src -I/data/data/com.termux/files/usr/include/python3.13 -c bindings/python/tree_sitter_typescript/binding.c -o build/temp.android-24-arm64_v8a-cpython-313/bindings/python/tree_sitter_typescript/binding.o -std=c11 -fvisibility=hidden
  clang -Wno-error=implicit-function-declaration -Wno-c2y-extensions -Wno-unused-variable -Wno-error=c2y-extensions -fvisibility=default -fPIC -DPy_LIMITED_API=0x03090000 -DPY_SSIZE_T_CLEAN -DTREE_SITTER_HIDE_SYMBOLS -Itypescript/src -I/data/data/com.termux/files/usr/include/python3.13 -c tsx/src/parser.c -o build/temp.android-24-arm64_v8a-cpython-313/tsx/src/parser.o -std=c11 -fvisibility=hidden
  tsx/src/parser.c:2929:14: error: unknown type name 'TSFieldMapSlice'
   2929 | static const TSFieldMapSlice ts_field_map_slices[PRODUCTION_ID_COUNT] = {
        |              ^
  1 error generated.
  error: command '/data/data/com.termux/files/usr/bin/clang' failed with exit code 1
  error: subprocess-exited-with-error
  

Why git clone + local pip install worked:

• I cloned tree-sitter-typescript repo → local pip install . used your local copy (likely with correct submodules or headers resolved)
• Build succeeded because the repo includes everything needed (parser generator, bindings) → no external header fetch needed
• Wheel built locally → pip happy

Why pip install docling failed initially:

• Pulled 0.23.2 (or HEAD) from PyPI/git → no Android wheel → source build → header missing → error

Now, as for over a year, works again:

Using cached docling-2.77.0-py3-none-any.whl (400 kB)
Using cached huggingface_hub-0.36.2-py3-none-any.whl (566 kB)
Using cached rapidocr-3.7.0-py3-none-any.whl (15.1 MB)
Using cached semchunk-2.2.2-py3-none-any.whl (10 kB)
Using cached transformers-4.57.6-py3-none-any.whl (12.0 MB)
Using cached mpire-2.10.2-py3-none-any.whl (272 kB)
Installing collected packages: tree-sitter-python, tree-sitter-javascript, tree-sitter-c, tree-sitter, pyclipper, mpire, rapidocr, pandas, huggingface_hub, semchunk, transformers, docling
  Attempting uninstall: pandas
    Found existing installation: pandas 3.0.1
    Uninstalling pandas-3.0.1:
      Successfully uninstalled pandas-3.0.1
  Attempting uninstall: huggingface_hub
    Found existing installation: huggingface_hub 1.6.0
    Uninstalling huggingface_hub-1.6.0:
      Successfully uninstalled huggingface_hub-1.6.0
  Attempting uninstall: transformers
    Found existing installation: transformers 5.3.0
    Uninstalling transformers-5.3.0:
      Successfully uninstalled transformers-5.3.0
  Attempting uninstall: docling
    Found existing installation: docling 2.76.0
    Uninstalling docling-2.76.0:
Successfully installed docling-2.77.0 huggingface_hub-0.36.2 mpire-2.10.2 pandas-2.3.3 pyclipper-1.4.0 rapidocr-3.7.0 semchunk-2.2.2 transformers-4.57.6 tree-sitter-0.25.2 tree-sitter-c-0.24.1 tree-sitter-javascript-0.25.0 tree-sitter-python-0.25.0