Update Mac (M1/M2) one click installers

h2oai · Mar 6, 2024 · a3cb88a · a3cb88a
1 parent dadcb9a
commit a3cb88a
Show file tree

Hide file tree

Showing 3 changed files with 47 additions and 14 deletions.
diff --git a/dev_installers/mac/README.md b/dev_installers/mac/README.md
@@ -17,12 +17,31 @@ This document provide the details to build one click installers for MacOS. To ma
 
 - Clone `h2ogpt` from https://github.com/h2oai/h2ogpt.git
 - Create conda environment and installer all required dependencies, consult [build_mac_installer.sh](build_mac_installer.sh) for more details.
-- Run below commands to build the installer
+- Run below commands to build the spec file for installer, replace the `--name` appropriately depending on whether building for CPU only or with MPS (GPU) support
     ```shell
     cd h2ogpt
-    pyinstaller ./dev_installers/mac/mac_run_app.py -D -w --name=h2ogpt-osx-m1-cpu-debug --hiddenimport=h2ogpt --collect-all=h2ogpt --noconfirm --recursive-copy-metadata=transformers --collect-data=langchain --collect-data=gradio_client --collect-all=gradio --path=${CONDA_PREFIX}/python3.10/site-packages --collect-all=sentencepiece --add-data=./Tesseract-OCR:Tesseract-OCR --add-data=./poppler:poppler
+    pyi-makespec mac_run_app.py -F --name=h2ogpt-osx-m1-cpu \
+      --hidden-import=h2ogpt \
+      --collect-all=h2ogpt \
+      --recursive-copy-metadata=transformers \
+      --collect-data=langchain \
+      --collect-data=gradio_client \
+      --collect-all=gradio \
+      --collect-all=sentencepiece \
+      --collect-all=gradio_pdf \
+      --collect-all=llama_cpp \
+      --collect-all=tiktoken_ext \
+      --add-data=../../Tesseract-OCR:Tesseract-OCR \
+      --add-data=../../poppler:poppler
     ```
-
+- Edit the `h2ogpt-osx-m1-cpu.spec` and/or `h2ogpt-osx-m1-gpu.spec` and add below code block to `Analysis()`, to explicitly tell PyInstaller to collect all `.py` modules from listed dependencies.
+    ```
+    module_collection_mode={
+        'gradio' : 'py',
+        'gradio_pdf' : 'py',
+    },
+    ```
+- Run `pyinstaller h2ogpt-osx-m1-cpu.spec` to build the installer.
 ### Deployment Mode
 
 - Clone `h2ogpt` from https://github.com/h2oai/h2ogpt.git

diff --git a/dev_installers/mac/build_mac_installer.sh b/dev_installers/mac/build_mac_installer.sh
@@ -6,9 +6,13 @@ then
     echo "conda could not be found, need conda to continue!"
     exit 1
 fi
-conda env remove -n h2ogpt-mac2
-conda create -n h2ogpt-mac2 python=3.10 rust -y
-conda activate h2ogpt-mac2
+
+# Remove old Tesseract and poppler deps
+rm -rf ./Tesseract-OCR poppler
+
+conda env remove -n h2ogpt-mac
+conda create -n h2ogpt-mac python=3.10 rust -y
+conda activate h2ogpt-mac
 
 pip install --upgrade pip
 python -m pip install --upgrade setuptools
@@ -27,21 +31,21 @@ pip install -r reqs_optional/requirements_optional_doctr.txt -c reqs_optional/re
 # Optional: for supporting unstructured package
 python -m nltk.downloader all
 
+# Required for CPU: LLaMa/GPT4All:
 # For MPS support
 if [ -z "$BUILD_MPS" ]
 then
     echo "BUILD_MPS is not set, skipping MPS specific configs..."
+    pip uninstall llama-cpp-python -y
+    CMAKE_ARGS="-DLLAMA_METAL=off" FORCE_CMAKE=1 pip install -r reqs_optional/requirements_optional_llamacpp_gpt4all.txt -c reqs_optional/reqs_constraints.txt --no-cache-dir
 else
     if [ "$BUILD_MPS" = "1" ]
     then
         echo "BUILD_MPS is set to 1, running MPS specific configs..."
         pip uninstall llama-cpp-python -y
-        export CMAKE_ARGS="-DLLAMA_METAL=on" FORCE_CMAKE=1
+        CMAKE_ARGS="-DLLAMA_METAL=on" FORCE_CMAKE=1 pip install -r reqs_optional/requirements_optional_llamacpp_gpt4all.txt -c reqs_optional/reqs_constraints.txt --no-cache-dir
     fi
 fi
-
-# Required for CPU: LLaMa/GPT4All:
-pip install -r reqs_optional/requirements_optional_gpt4all.txt -c reqs_optional/reqs_constraints.txt --no-cache-dir
 pip install librosa -c reqs_optional/reqs_constraints.txt
 
 # Install PyInstaller
@@ -58,7 +62,7 @@ make clean dist
 pip install ./dist/h2ogpt*.whl
 
 # Build Mac Installer
-# below command is used to build current .spec file replace it whenever use new configs
+# below command is used to build current .spec file from project root, replace it whenever use new configs
 #pyi-makespec mac_run_app.py -F --name=h2ogpt-osx-m1-cpu \
 #  --hidden-import=h2ogpt \
 #  --collect-all=h2ogpt \
@@ -70,8 +74,8 @@ pip install ./dist/h2ogpt*.whl
 #  --collect-all=gradio_pdf \
 #  --collect-all=llama_cpp \
 #  --collect-all=tiktoken_ext \
-#  --add-data=./Tesseract-OCR:Tesseract-OCR \
-#  --add-data=./poppler:poppler
+#  --add-data=../../Tesseract-OCR:Tesseract-OCR \
+#  --add-data=../../poppler:poppler
 
 # add below argument to Analysis() call in h2ogpt-osx-m1-cpu.spec file
 #module_collection_mode={

diff --git a/dev_installers/mac/h2ogpt-osx-m1-gpu.spec b/dev_installers/mac/h2ogpt-osx-m1-gpu.spec
@@ -15,10 +15,16 @@ tmp_ret = collect_all('gradio')
 datas += tmp_ret[0]; binaries += tmp_ret[1]; hiddenimports += tmp_ret[2]
 tmp_ret = collect_all('sentencepiece')
 datas += tmp_ret[0]; binaries += tmp_ret[1]; hiddenimports += tmp_ret[2]
+tmp_ret = collect_all('gradio_pdf')
+datas += tmp_ret[0]; binaries += tmp_ret[1]; hiddenimports += tmp_ret[2]
+tmp_ret = collect_all('llama_cpp')
+datas += tmp_ret[0]; binaries += tmp_ret[1]; hiddenimports += tmp_ret[2]
+tmp_ret = collect_all('tiktoken_ext')
+datas += tmp_ret[0]; binaries += tmp_ret[1]; hiddenimports += tmp_ret[2]
 
 
 a = Analysis(
-    ['./mac_run_app.py'],
+    ['mac_run_app.py'],
     pathex=[],
     binaries=binaries,
     datas=datas,
@@ -28,6 +34,10 @@ a = Analysis(
     runtime_hooks=[],
     excludes=[],
     noarchive=False,
+    module_collection_mode={
+        'gradio' : 'py',
+        'gradio_pdf' : 'py',
+    },
 )
 pyz = PYZ(a.pure)