bulk-chain 1.1.0__py3-none-any.whl → 1.2.1__py3-none-any.whl

bulk_chain/api.py

@@ -3,6 +3,7 @@ import collections
 import logging
 import os
 from itertools import chain
+from types import AsyncGeneratorType
 
 from bulk_chain.core.llm_base import BaseLM
 from bulk_chain.core.service_asyncio import AsyncioService
@@ -16,8 +17,8 @@ from bulk_chain.core.utils import attempt_wrapper
 
 INFER_MODES = {
     "single": lambda llm, batch, **kwargs: [llm.ask(prompt) for prompt in batch],
+    "batch": lambda llm, batch, **kwargs: llm.ask_batch(batch),
     "single_stream": lambda llm, batch, **kwargs: [llm.ask_stream(prompt) for prompt in batch],
-    "batch": lambda llm, batch, **kwargs: llm.ask(batch),
     "batch_async": lambda llm, batch, **kwargs: AsyncioService.run_tasks(
         batch=batch, async_handler=llm.ask_async, event_loop=kwargs.get("event_loop")
     ),
@@ -69,6 +70,9 @@ def __handle_gen(handle, batch, event_loop):
         elif isinstance(entry, collections.abc.Iterable):
             for chunk in map(lambda item: str(item), entry):
                 yield chunk
+        elif isinstance(entry, AsyncGeneratorType):
+            for chunk in AsyncioService.async_gen_to_iter(entry, loop=event_loop):
+                yield str(chunk)
         else:
             raise Exception(f"Non supported type `{type(entry)}` for handling output from batch")
 
@@ -85,11 +89,14 @@ def _iter_chunks(p_column, batch_content_it, **kwargs):
         yield ind_in_batch, chunk
 
 
-def _infer_batch(batch, batch_ind, schema, return_mode, cols=None, **kwargs):
+def _column_ordered_chunks_iter(batch, schema, cols=None, keep_prompts=True, **kwargs):
+    """
+    NOTE: we populate `batch` content automatically
+    """
     assert (isinstance(batch, list))
 
     if len(batch) == 0:
-        return batch
+        return
 
     if cols is None:
         first_item = batch[0]
@@ -112,33 +119,66 @@ def _infer_batch(batch, batch_ind, schema, return_mode, cols=None, **kwargs):
             for ind_in_batch, chunk in content_it:
                 # Append batch.
                 batch[ind_in_batch][c].append(chunk)
-                # Returning (optional).
-                if return_mode == "chunk":
-                    global_ind = batch_ind * len(batch) + ind_in_batch
-                    yield [global_ind, c, chunk]
+                yield [ind_in_batch, c, chunk]
 
             # Convert content to string.
             for item in batch:
                 item[c] = "".join(item[c])
 
-    if return_mode == "record":
+    if not keep_prompts:
+        for batch_item in batch:
+            for key in list(batch_item.keys()):
+                prompt_col = SchemaService.col_to_prompt(col_name=key, prompt_data=batch_item)
+                if prompt_col in batch_item:
+                    del batch_item[prompt_col]
+
+
+def _infer_batch(return_type, batch, batch_ind, **kwargs):
+    assert (return_type in ["batch", "chunk", "record"])
+
+    # Filling batch with inference content.
+    for ind_in_batch, column, chunk in _column_ordered_chunks_iter(batch=batch, **kwargs):
+        if return_type == "chunk":
+            global_ind = batch_ind * len(batch) + ind_in_batch
+            yield [global_ind, column, chunk]
+
+    if return_type == "record":
         for record in batch:
             yield record
 
-    if return_mode == "batch":
+    if return_type == "batch":
         yield batch
 
 
+def get_infer_mode(stream, batch_size, async_mode):
+    if not stream and batch_size == 1:
+        return 'single', 'record'
+    elif not stream and batch_size > 1:
+        if async_mode:
+            return 'batch_async', 'batch'
+        else:
+            return 'batch', 'batch'
+    elif stream and batch_size == 1:
+        return 'single_stream', 'chunk'
+    elif stream and batch_size > 1:
+        return 'batch_stream_async', 'chunk'
+
+    raise ValueError(f"Invalid combination of stream and batch_size: {stream}, {batch_size}")
+
+
 def iter_content(input_dicts_it, llm, schema, batch_size=1, limit_prompt=None,
-                 infer_mode="batch", return_mode="batch", attempts=1, event_loop=None,
-                 **kwargs):
+                 stream=False, async_mode=False, attempts=1, event_loop=None,
+                 handle_missed_value_func=lambda *_: None, **kwargs):
     """ This method represent Python API aimed at application of `llm` towards
         iterator of input_dicts via cache_target that refers to the SQLite using
         the given `schema`
     """
-    assert (infer_mode in INFER_MODES.keys())
-    assert (return_mode in ["batch", "chunk", "record"])
     assert (isinstance(llm, BaseLM))
+    assert (isinstance(batch_size, int) and batch_size > 0)
+    assert (isinstance(async_mode, bool)) 
+
+    infer_type, return_type = get_infer_mode(stream=stream, batch_size=batch_size, async_mode=async_mode)
+    infer_mode = INFER_MODES[infer_type]
 
     # Setup event loop.
     event_loop = asyncio.get_event_loop_policy().get_event_loop() \
@@ -149,13 +189,15 @@ def iter_content(input_dicts_it, llm, schema, batch_size=1, limit_prompt=None,
         schema = JsonService.read(schema)
     if isinstance(schema, dict):
         schema = SchemaService(json_data=schema)
+    if isinstance(schema, list):
+        schema = SchemaService(json_data={"schema": schema})
 
     prompts_it = map(
         lambda data: DictionaryService.custom_update(src_dict=dict(data), other_dict=schema.cot_args),
         input_dicts_it
     )
 
-    handle_batch_func = lambda batch, **handle_kwargs: INFER_MODES[infer_mode](
+    handle_batch_func = lambda batch, **handle_kwargs: infer_mode(
         llm,
         DataService.limit_prompts(batch, limit=limit_prompt),
         **handle_kwargs
@@ -172,12 +214,13 @@ def iter_content(input_dicts_it, llm, schema, batch_size=1, limit_prompt=None,
                                       logger=logger)
         handle_batch_func = attempt_dec(handle_batch_func)
 
-    content_it = (_infer_batch(batch=batch,
+    kwargs["handle_missed_value_func"] = handle_missed_value_func
+
+    content_it = (_infer_batch(return_type=return_type,
+                               batch=batch,
                                batch_ind=batch_ind,
                                infer_mode=infer_mode,
                                handle_batch_func=handle_batch_func,
-                               handle_missed_value_func=lambda *_: None,
-                               return_mode=return_mode,
                                schema=schema,
                                event_loop=event_loop,
                                **kwargs)

bulk_chain/core/llm_base.py

@@ -3,12 +3,17 @@ class BaseLM(object):
     def __init__(self, **kwargs):
         pass
 
-    def ask(self, content):
+    def ask(self, prompt):
         """ Assumes to return str.
         """
         raise NotImplemented()
 
-    def ask_stream(self, content):
+    def ask_batch(self, batch):
+        """ Assumes to return generator.
+        """
+        raise NotImplemented()
+
+    def ask_stream(self, prompt):
         """ Assumes to return generator.
         """
         raise NotImplemented()
@@ -18,7 +23,7 @@ class BaseLM(object):
         """
         raise NotImplemented()
 
-    async def ask_stream_async(self, batch):
+    async def ask_stream_async(self, prompt):
         """ Assumes to return AsyncGenerator.
         """
         raise NotImplemented()

bulk_chain/core/service_schema.py

@@ -9,6 +9,10 @@ class SchemaService(object):
         prompt_schema = {"schema": [{"prompt": prompt, "out": "response", "in": "prompt"}]}
         return cls(prompt_schema)
 
+    @staticmethod
+    def col_to_prompt(col_name, prompt_data):
+        return col_name + "_prompt" if "in" not in prompt_data else prompt_data["in"]
+
     @staticmethod
     def __init_schema(prompts):
 
@@ -19,7 +23,7 @@ class SchemaService(object):
 
         for prompt in prompts:
             r_col_name = prompt["out"]
-            p_col_name = r_col_name + "_prompt" if "in" not in prompt else prompt["in"]
+            p_col_name = SchemaService.col_to_prompt(col_name=r_col_name, prompt_data=prompt)
 
             assert r_col_name not in schema_r2p, f"`{r_col_name}` has been already declared!"
             assert p_col_name not in schema_p2r, f"`{p_col_name}` has been already declared!"

bulk_chain/core/utils.py

@@ -1,3 +1,4 @@
+import ast
 import importlib
 import logging
 import sys
@@ -35,18 +36,30 @@ def find_by_prefix(d, key):
     return d[matches[0]]
 
 
+def check_is_param_name(param_name):
+    return param_name.replace("_", "").isalpha()
+
+
 def iter_params(text):
     assert(isinstance(text, str))
     beg = 0
     while beg < len(text):
+        print(beg)
         try:
             pb = text.index('{', beg)
         except ValueError:
             break
-        pe = text.index('}', beg+1)
-        # Yield argument.
-        yield text[pb+1:pe]
-        beg = pe+1
+        pe = text.index('}', pb+1)
+        param_name = text[pb + 1:pe]
+
+        # Check parameter validity.
+        if not check_is_param_name(param_name):
+            beg = pb + 1
+            continue
+
+        # Passing.
+        yield param_name
+        beg = pe + 1
 
 
 def auto_import(name, is_class=False):
@@ -61,6 +74,17 @@ def auto_import(name, is_class=False):
     return m() if is_class else m
 
 
+def get_class_name(file_path):
+    with open(file_path, 'r') as f:
+        tree = ast.parse(f.read(), filename=file_path)
+
+    for node in ast.walk(tree):
+        if isinstance(node, ast.ClassDef):
+            return node.name
+
+    return None
+
+
 def dynamic_init(class_filepath, class_name=None):
 
     # Registering path.
@@ -75,7 +99,7 @@ def dynamic_init(class_filepath, class_name=None):
         class_filename = class_filename[:-len(".py")]
 
     # Loading library.
-    class_name = class_path_list[-1].title() if class_name is None else class_name
+    class_name = get_class_name(class_filepath) if class_name is None else class_name
     class_path = ".".join([class_filename, class_name])
     logger.info(f"Dynamic loading for the file and class `{class_path}`")
     cls = auto_import(class_path, is_class=False)

{bulk_chain-1.1.0.dist-info → bulk_chain-1.2.1.dist-info}/METADATA

@@ -1,12 +1,13 @@
 Metadata-Version: 2.1
-Name: bulk_chain
-Version: 1.1.0
+Name: bulk-chain
+Version: 1.2.1
 Summary: A lightweight, no-strings-attached Chain-of-Thought framework for your LLM, ensuring reliable results for bulk input requests.
 Home-page: https://github.com/nicolay-r/bulk-chain
 Author: Nicolay Rusnachenko
 Author-email: rusnicolay@gmail.com
 License: MIT License
 Keywords: natural language processing,chain-of-thought,reasoning
+Platform: UNKNOWN
 Classifier: Programming Language :: Python
 Classifier: Programming Language :: Python :: 3.9
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
@@ -14,9 +15,8 @@ Classifier: Topic :: Scientific/Engineering :: Information Analysis
 Classifier: Topic :: Text Processing :: Linguistic
 Requires-Python: >=3.6
 Description-Content-Type: text/markdown
-License-File: LICENSE
 
-# bulk-chain 1.1.0
+# bulk-chain 1.2.1
 ![](https://img.shields.io/badge/Python-3.9-brightgreen.svg)
 [![](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/nicolay-r/bulk-chain/blob/master/bulk_chain_tutorial.ipynb)
 [![twitter](https://img.shields.io/twitter/url/https/shields.io.svg?style=social)](https://x.com/nicolayr_/status/1847969224636961033)
@@ -55,24 +55,17 @@ pip install git+https://github.com/nicolay-r/bulk-chain@master
 
 ## Chain-of-Thought Schema
 
-To declare Chain-of-Though (CoT) schema, this project exploits `JSON` format.
-This format adopts `name` field for declaring a name and `schema` is a list of CoT instructions for the Large Language Model.
-
-Each step represents a dictionary with `prompt` and `out` keys that corresponds to the input prompt and output variable name respectively.
-All the variable names are expected to be mentioned in `{}`.
-
-Below, is an example on how to declare your own schema:
+To declare Chain-of-Though (CoT) schema we use `JSON` format. 
+The field `schema` is a list of CoT instructions for the Large Language Model.
+Each item of the list represent a dictionary with `prompt` and `out` keys that corresponds to the input prompt and output variable name respectively.
+All the variable names should be mentioned in `{}`.
 
+**Example**:
 ```python
-{
-"name": "schema-name",
-"schema": [
-    {"prompt": "Given the question '{text}', let's think step-by-step.", 
-     "out": "steps"},
-    {"prompt": "For the question '{text}' the reasoining steps are '{steps}'. what would be an answer?", 
-     "out":  "answer"},
+[
+  {"prompt": "extract topic: {text}", "out": "topic"},
+  {"prompt": "extract subject: {text}", "out": "subject"},
 ]
-}
 ```
 
 # Usage
@@ -94,25 +87,50 @@ from bulk_chain.api import iter_content
 
 content_it = iter_content(
     # 1. Your schema.              
-    schema="YOUR_SCHEMA.json",
+    schema=[
+      {"prompt": "extract topic: {text}", "out": "topic" },
+      {"prompt": "extract subject: {text}", "out": "subject"},
+    ],
     # 2. Your third-party model implementation.
-    llm=dynamic_init(class_filepath="replicate_104.py", class_name="Replicate")(api_token="<API-KEY>"),
-    # 3. Customize your inference and result providing modes: 
-    infer_mode="batch_async", 
-    return_mode="batch",
-    # 4. Your iterator of dictionaries
-    input_dicts_it=YOUR_DATA_IT,
+    llm=dynamic_init(class_filepath="replicate_104.py")(
+       api_token="<API-KEY>",
+       model_name="meta/meta-llama-3-70b-instruct"),
+    # 3. Toggle streaming if needed
+    stream=False,
+    # 4. Toggle Async API mode usage.
+    async_mode=True,
+    # 5. Batch size.
+    batch_size=10,
+    # 6. Your iterator of dictionaries
+    input_dicts_it=[
+        # Example of data ...
+        { "text": "Rocks are hard" },
+        { "text": "Water is wet" },
+        { "text": "Fire is hot" }
+    ],
 )
-    
-for content in content_it:
-    # Handle your LLM responses here ...
+
+for batch in content_it:
+   for entry in batch:
+      print(entry)
 ```
 
+Outputs entries represent texts augmented with `topic` and `subject`:
+```jsonl
+{'text': 'Rocks are hard', 'topic': 'The topic is: Geology/Rocks', 'subject': 'The subject is: "Rocks"'}
+{'text': 'Water is wet', 'topic': 'The topic is: Properties of Water', 'subject': 'The subject is: Water'}
+{'text': 'Fire is hot', 'topic': 'The topic is: Temperature/Properties of Fire', 'subject': 'The subject is: "Fire"'}
+```
 
-# Embed your LLM
+# API
 
-All you have to do is to implement `BaseLM` class, that includes:
-* `__init__` -- for setting up *batching mode support* and (optional) *model name*;
-* `ask(prompt)` -- infer your model with the given `prompt`.
+| Method               | Mode       | Description                                                         |
+|----------------------|------------|---------------------------------------------------------------------|
+| `ask(prompt)`        | Sync       | Infers the model with a single prompt.                              |
+| `ask_stream(prompt)` | Sync       | Returns a generator that yields chunks of the inferred result.      |
+| `ask_async(prompt)`  | Async      | Asynchronously infers the model with a single prompt.               |
+| `ask_stream_async(prompt)` | Async | Asynchronously returns a generator of result chunks of the inferred result.          |
 
 See examples with models [at nlp-thirdgate 🌌](https://github.com/nicolay-r/nlp-thirdgate?tab=readme-ov-file#llm).
+
+

bulk_chain-1.2.1.dist-info/RECORD

@@ -0,0 +1,16 @@
+bulk_chain/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+bulk_chain/api.py,sha256=bLZXdp58i6LDayZQxRBxsFK4lVT8cZZn1uOY0iaZ5TE,8500
+bulk_chain/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+bulk_chain/core/llm_base.py,sha256=H2KmCqChKp9sKOkROE-4zjMRCxizT9xWvNZSF22HeFU,673
+bulk_chain/core/service_asyncio.py,sha256=S-D4K3LBa3noKTm0tXazluYVI8cBgN1IB6v6MFoMyNQ,1972
+bulk_chain/core/service_batch.py,sha256=lWmjO0aU6h2rmfx_kGmNqt0Rdeaf2a4Dn5VyfKFkfDs,1033
+bulk_chain/core/service_data.py,sha256=OWWHHnr_plwxYTxLuvMrhEc1PbSx-XC3rbFzV0hy3vk,1107
+bulk_chain/core/service_dict.py,sha256=lAghLU-3V3xYGv5BTA327Qcw8UJYmgQRMFdggzlrUgo,383
+bulk_chain/core/service_json.py,sha256=6o1xM_8c9QEjH9Q3qEmJylU9nahfRXhUd5sFF2dGJwo,182
+bulk_chain/core/service_schema.py,sha256=YAsdm3N2G4-eTpeJazg4Y-KQ2w9bEPpqreVl8a-M7H0,1311
+bulk_chain/core/utils.py,sha256=hml0zLmnZe865gvc1CagEzRE19Gdh1pF8kx_KueDY3A,3667
+bulk_chain-1.2.1.dist-info/LICENSE,sha256=VF9SjNpwwSSFEY_eP_8A1ocDCrbwfjI1pZexXdCkOwo,1076
+bulk_chain-1.2.1.dist-info/METADATA,sha256=xx1vcG6wkHzh_Ga3iZJV3MBdR97RBGpCf7JO5_lonN0,5339
+bulk_chain-1.2.1.dist-info/WHEEL,sha256=g4nMs7d-Xl9-xC9XovUrsDHGXt-FT0E17Yqo92DEfvY,92
+bulk_chain-1.2.1.dist-info/top_level.txt,sha256=Hxq_wyH-GDXKBaA63UfBIiMJO2eCHJG5EOrXDphpeB4,11
+bulk_chain-1.2.1.dist-info/RECORD,,

{bulk_chain-1.1.0.dist-info → bulk_chain-1.2.1.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.1.0)
+Generator: bdist_wheel (0.34.2)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

bulk_chain-1.1.0.dist-info/RECORD

@@ -1,16 +0,0 @@
-bulk_chain/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-bulk_chain/api.py,sha256=gPGjaHYIn2Ewn6yXIXER-CM5SgXQ3ZJH-SdRyaPDOo0,6890
-bulk_chain/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-bulk_chain/core/llm_base.py,sha256=aa73TGW03yLXMHY4b_1NgquRvP0CzH8IWZkcFPABFUg,557
-bulk_chain/core/service_asyncio.py,sha256=S-D4K3LBa3noKTm0tXazluYVI8cBgN1IB6v6MFoMyNQ,1972
-bulk_chain/core/service_batch.py,sha256=lWmjO0aU6h2rmfx_kGmNqt0Rdeaf2a4Dn5VyfKFkfDs,1033
-bulk_chain/core/service_data.py,sha256=OWWHHnr_plwxYTxLuvMrhEc1PbSx-XC3rbFzV0hy3vk,1107
-bulk_chain/core/service_dict.py,sha256=lAghLU-3V3xYGv5BTA327Qcw8UJYmgQRMFdggzlrUgo,383
-bulk_chain/core/service_json.py,sha256=6o1xM_8c9QEjH9Q3qEmJylU9nahfRXhUd5sFF2dGJwo,182
-bulk_chain/core/service_schema.py,sha256=KIP4n0Tz2h1i7SIMGhgAhoiCgUFXOT1rzMt38yACS2U,1154
-bulk_chain/core/utils.py,sha256=tp1FJQBmJt-3QmG7B0hyJNTFyg_8BwTTdl8xTxSgNDk,3140
-bulk_chain-1.1.0.dist-info/LICENSE,sha256=VF9SjNpwwSSFEY_eP_8A1ocDCrbwfjI1pZexXdCkOwo,1076
-bulk_chain-1.1.0.dist-info/METADATA,sha256=EheCGDisKF0TwmzJfnDxW-rgsDVPNpCYGOvuaDn91tw,4428
-bulk_chain-1.1.0.dist-info/WHEEL,sha256=GV9aMThwP_4oNCtvEC2ec3qUYutgWeAzklro_0m4WJQ,91
-bulk_chain-1.1.0.dist-info/top_level.txt,sha256=Hxq_wyH-GDXKBaA63UfBIiMJO2eCHJG5EOrXDphpeB4,11
-bulk_chain-1.1.0.dist-info/RECORD,,