spark-nlp 2.6.3rc1__py2.py3-none-any.whl → 6.2.1__py2.py3-none-any.whl

sparknlp/reader/pdf_to_text.py

@@ -0,0 +1,190 @@
+#  Copyright 2017-2025 John Snow Labs
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+from pyspark import keyword_only
+from pyspark.ml.param import Param, Params, TypeConverters
+from pyspark.ml.param.shared import HasInputCol, HasOutputCol
+from pyspark.ml.util import JavaMLReadable, JavaMLWritable
+from pyspark.ml.wrapper import JavaTransformer
+
+from sparknlp.reader.enums import TextStripperType
+
+
+class PdfToText(JavaTransformer, HasInputCol, HasOutputCol,
+                JavaMLReadable, JavaMLWritable):
+    """
+    Extract text from PDF documents as either a single string or multiple strings per page.
+    Input is a column with binary content of PDF files. Output is a column with extracted text,
+    with options to include page numbers or split pages.
+
+    Parameters
+    ----------
+    pageNumCol : str, optional
+        Page number output column name.
+    partitionNum : int, optional
+        Number of partitions (default is 0).
+    storeSplittedPdf : bool, optional
+        Whether to store content of split PDFs (default is False).
+    splitPage : bool, optional
+        Enable/disable splitting per page (default is True).
+    onlyPageNum : bool, optional
+        Whether to extract only page numbers (default is False).
+    textStripper : str or TextStripperType, optional
+        Defines layout and formatting type.
+    sort : bool, optional
+        Enable/disable sorting content per page (default is False).
+
+    Examples
+    --------
+    >>> import sparknlp
+    >>> from sparknlp.reader import *
+    >>> from pyspark.ml import Pipeline
+    >>> pdf_path = "Documents/files/pdf"
+    >>> data_frame = spark.read.format("binaryFile").load(pdf_path)
+    >>> pdf_to_text = PdfToText().setStoreSplittedPdf(True)
+    >>> pipeline = Pipeline(stages=[pdf_to_text])
+    >>> pipeline_model = pipeline.fit(data_frame)
+    >>> pdf_df = pipeline_model.transform(data_frame)
+    >>> pdf_df.show()
+    +--------------------+--------------------+
+    |                path|    modificationTime|
+    +--------------------+--------------------+
+    |file:/Users/paula...|2025-05-15 11:33:...|
+    |file:/Users/paula...|2025-05-15 11:33:...|
+    +--------------------+--------------------+
+    >>> pdf_df.printSchema()
+    root
+     |-- path: string (nullable = true)
+     |-- modificationTime: timestamp (nullable = true)
+     |-- length: long (nullable = true)
+     |-- text: string (nullable = true)
+     |-- height_dimension: integer (nullable = true)
+     |-- width_dimension: integer (nullable = true)
+     |-- content: binary (nullable = true)
+     |-- exception: string (nullable = true)
+     |-- pagenum: integer (nullable = true)
+    """
+    pageNumCol = Param(Params._dummy(), "pageNumCol",
+                       "Page number output column name.",
+                       typeConverter=TypeConverters.toString)
+
+    partitionNum = Param(Params._dummy(), "partitionNum",
+                         "Number of partitions.",
+                         typeConverter=TypeConverters.toInt)
+
+    storeSplittedPdf = Param(Params._dummy(), "storeSplittedPdf",
+                             "Force to store splitted pdf.",
+                             typeConverter=TypeConverters.toBoolean)
+
+    splitPage = Param(Params._dummy(), "splitPage",
+                      "Param for enable/disable splitting document per page",
+                      typeConverter=TypeConverters.toBoolean)
+
+    textStripper = Param(Params._dummy(), "textStripper",
+                         "Text stripper type used for output layout and formatting",
+                         typeConverter=TypeConverters.toString)
+
+    sort = Param(Params._dummy(), "sort",
+                 "Param for enable/disable sort lines",
+                 typeConverter=TypeConverters.toBoolean)
+
+    onlyPageNum = Param(Params._dummy(), "onlyPageNum",
+                        "Force to extract only number of pages",
+                        typeConverter=TypeConverters.toBoolean)
+
+    extractCoordinates = Param(Params._dummy(), "extractCoordinates",
+                               "Force extract coordinates of text.",
+                               typeConverter=TypeConverters.toBoolean)
+
+    normalizeLigatures = Param(Params._dummy(), "normalizeLigatures",
+                               "Whether to convert ligature chars such as 'fl' into its corresponding chars (e.g., {'f', 'l'}).",
+                               typeConverter=TypeConverters.toBoolean)
+
+    @keyword_only
+    def __init__(self):
+        """
+        __init__(self)
+        """
+        super(PdfToText, self).__init__()
+        self._java_obj = self._new_java_obj("com.johnsnowlabs.reader.PdfToText", self.uid)
+
+    def setInputCol(self, value):
+        """
+        Sets the value of :py:attr:`inputCol`.
+        """
+        return self._set(inputCol=value)
+
+    def setOutputCol(self, value):
+        """
+        Sets the value of :py:attr:`outputCol`.
+        """
+        return self._set(outputCol=value)
+
+    def setPageNumCol(self, value):
+        """
+        Sets the value of :py:attr:`pageNumCol`.
+        """
+        return self._set(pageNumCol=value)
+
+    def setPartitionNum(self, value):
+        """
+        Sets the value of :py:attr:`partitionNum`.
+        """
+        return self._set(partitionNum=value)
+
+    def setStoreSplittedPdf(self, value):
+        """
+        Sets the value of :py:attr:`storeSplittedPdf`.
+        """
+        return self._set(storeSplittedPdf=value)
+
+    def setSplitPage(self, value):
+        """
+        Sets the value of :py:attr:`splitPage`.
+        """
+        return self._set(splitPage=value)
+
+    def setOnlyPageNum(self, value):
+        """
+        Sets the value of :py:attr:`onlyPageNum`.
+        """
+        return self._set(onlyPageNum=value)
+
+    def setTextStripper(self, value):
+        """
+        Sets the value of :py:attr:`textStripper`.
+        """
+        if isinstance(value, TextStripperType):
+            value = value.value
+        if value not in [i.value for i in TextStripperType]:
+            type_value = type(value)
+            raise ValueError(f"Param textStripper must be a 'TextStripperType' enum but got {type_value}.")
+        return self._set(textStripper=str(value))
+
+    def setSort(self, value):
+        """
+        Sets the value of :py:attr:`sort`.
+        """
+        return self._set(sort=value)
+
+    def setExtractCoordinates(self, value):
+        """
+        Sets the value of :py:attr:`extractCoordinates`.
+        """
+        return self._set(extractCoordinates=value)
+
+    def setNormalizeLigatures(self, value):
+        """
+        Sets the value of :py:attr:`normalizeLigatures`.
+        """
+        return self._set(normalizeLigatures=value)

sparknlp/reader/reader2doc.py

@@ -0,0 +1,124 @@
+#  Copyright 2017-2025 John Snow Labs
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+from pyspark import keyword_only
+
+from sparknlp.common import AnnotatorType
+from sparknlp.internal import AnnotatorTransformer
+from sparknlp.partition.partition_properties import *
+
+
+class Reader2Doc(
+    AnnotatorTransformer,
+    HasReaderProperties,
+    HasHTMLReaderProperties,
+    HasEmailReaderProperties,
+    HasExcelReaderProperties,
+    HasPowerPointProperties,
+    HasTextReaderProperties
+):
+    """
+    The Reader2Doc annotator allows you to use reading files more smoothly within existing
+    Spark NLP workflows, enabling seamless reuse of your pipelines.
+
+    Reader2Doc can be used for extracting structured content from various document types
+    using Spark NLP readers. It supports reading from many file types and returns parsed
+    output as a structured Spark DataFrame.
+
+    Supported formats include:
+
+    - Plain text
+    - HTML
+    - Word (.doc/.docx)
+    - Excel (.xls/.xlsx)
+    - PowerPoint (.ppt/.pptx)
+    - Email files (.eml, .msg)
+    - PDFs
+
+    Examples
+    --------
+    >>> from johnsnowlabs.reader import Reader2Doc
+    >>> from johnsnowlabs.nlp.base import DocumentAssembler
+    >>> from pyspark.ml import Pipeline
+    >>> # Initialize Reader2Doc for PDF files
+    >>> reader2doc = Reader2Doc() \\
+    ...     .setContentType("application/pdf") \\
+    ...     .setContentPath(f"{pdf_directory}/")
+    >>> # Build the pipeline with the Reader2Doc stage
+    >>> pipeline = Pipeline(stages=[reader2doc])
+    >>> # Fit the pipeline to an empty DataFrame
+    >>> pipeline_model = pipeline.fit(empty_data_set)
+    >>> result_df = pipeline_model.transform(empty_data_set)
+    >>> # Show the resulting DataFrame
+    >>> result_df.show()
+    +------------------------------------------------------------------------------------------------------------------------------------+
+    |document                                                                                                                            |
+    +------------------------------------------------------------------------------------------------------------------------------------+
+    |[{'document', 0, 14, 'This is a Title', {'pageNumber': 1, 'elementType': 'Title', 'fileName': 'pdf-title.pdf'}, []}]               |
+    |[{'document', 15, 38, 'This is a narrative text', {'pageNumber': 1, 'elementType': 'NarrativeText', 'fileName': 'pdf-title.pdf'}, []}]|
+    |[{'document', 39, 68, 'This is another narrative text', {'pageNumber': 1, 'elementType': 'NarrativeText', 'fileName': 'pdf-title.pdf'}, []}]|
+    +------------------------------------------------------------------------------------------------------------------------------------+
+    """
+
+    name = "Reader2Doc"
+
+    outputAnnotatorType = AnnotatorType.DOCUMENT
+
+    excludeNonText = Param(
+        Params._dummy(),
+        "excludeNonText",
+        "Whether to exclude non-text content from the output. Default is False.",
+        typeConverter=TypeConverters.toBoolean
+    )
+
+    def setExcludeNonText(self, value):
+        """Sets whether to exclude non-text content from the output.
+
+        Parameters
+        ----------
+        value : bool
+            Whether to exclude non-text content from the output. Default is False.
+        """
+        return self._set(excludeNonText=value)
+
+    joinString = Param(
+        Params._dummy(),
+        "joinString",
+        "If outputAsDocument is true, specifies the string used to join elements into a single document.",
+        typeConverter=TypeConverters.toString
+    )
+
+    def setJoinString(self, value):
+        """
+        If outputAsDocument is true, specifies the string used to join elements into a single
+        """
+        return self._set(joinString=value)
+
+    @keyword_only
+    def __init__(self):
+        super(Reader2Doc, self).__init__(classname="com.johnsnowlabs.reader.Reader2Doc")
+        self._setDefault(
+            outputCol="document",
+            explodeDocs=False,
+            contentType="",
+            flattenOutput=False,
+            outputAsDocument=True,
+            outputFormat="plain-text",
+            excludeNonText=False,
+            joinString="\n"
+        )
+
+    @keyword_only
+    def setParams(self):
+        kwargs = self._input_kwargs
+        return self._set(**kwargs)

sparknlp/reader/reader2image.py

@@ -0,0 +1,136 @@
+#  Copyright 2017-2025 John Snow Labs
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+from pyspark import keyword_only
+from pyspark.ml.param import TypeConverters, Params, Param
+
+from sparknlp.common import AnnotatorType
+from sparknlp.internal import AnnotatorTransformer
+from sparknlp.partition.partition_properties import *
+
+class Reader2Image(
+    AnnotatorTransformer,
+    HasReaderProperties,
+    HasHTMLReaderProperties,
+    HasPdfProperties
+):
+    """
+    The Reader2Image annotator allows you to use the reading files with images more smoothly within existing
+    Spark NLP workflows, enabling seamless reuse of your pipelines. Reader2Image can be used for
+    extracting structured image content from various document types using Spark NLP readers. It supports
+    reading from many file types and returns parsed output as a structured Spark DataFrame.
+
+    Supported formats include HTML and Markdown.
+
+    == Example ==
+    This example demonstrates how to load HTML files with images and process them into a structured
+    Spark DataFrame using Reader2Image.
+
+    Expected output:
+    +-------------------+--------------------+
+    |           fileName|               image|
+    +-------------------+--------------------+
+    |example-images.html|[{image, example-...|
+    |example-images.html|[{image, example-...|
+    +-------------------+--------------------+
+
+    Schema:
+    root
+     |-- fileName: string (nullable = true)
+     |-- image: array (nullable = false)
+     |    |-- element: struct (containsNull = true)
+     |    |    |-- annotatorType: string (nullable = true)
+     |    |    |-- origin: string (nullable = true)
+     |    |    |-- height: integer (nullable = false)
+     |    |    |-- width: integer (nullable = false)
+     |    |    |-- nChannels: integer (nullable = false)
+     |    |    |-- mode: integer (nullable = false)
+     |    |    |-- result: binary (nullable = true)
+     |    |    |-- metadata: map (nullable = true)
+     |    |    |    |-- key: string
+     |    |    |    |-- value: string (valueContainsNull = true)
+     |    |    |-- text: string (nullable = true)
+    """
+
+    name = "Reader2Image"
+    outputAnnotatorType = AnnotatorType.IMAGE
+
+    userMessage = Param(
+        Params._dummy(),
+        "userMessage",
+        "Custom user message.",
+        typeConverter=TypeConverters.toString
+    )
+
+    promptTemplate = Param(
+        Params._dummy(),
+        "promptTemplate",
+        "Format of the output prompt.",
+        typeConverter=TypeConverters.toString
+    )
+
+    customPromptTemplate = Param(
+        Params._dummy(),
+        "customPromptTemplate",
+        "Custom prompt template for image models.",
+        typeConverter=TypeConverters.toString
+    )
+
+    @keyword_only
+    def __init__(self):
+        super(Reader2Image, self).__init__(classname="com.johnsnowlabs.reader.Reader2Image")
+        self._setDefault(
+            contentType="",
+            outputFormat="image",
+            explodeDocs=True,
+            userMessage="Describe this image",
+            promptTemplate="qwen2vl-chat",
+            readAsImage=True,
+            customPromptTemplate="",
+            ignoreExceptions=True
+        )
+
+    @keyword_only
+    def setParams(self):
+        kwargs = self._input_kwargs
+        return self._set(**kwargs)
+
+    def setUserMessage(self, value: str):
+        """Sets custom user message.
+
+        Parameters
+        ----------
+        value : str
+            Custom user message to include.
+        """
+        return self._set(userMessage=value)
+
+    def setPromptTemplate(self, value: str):
+        """Sets format of the output prompt.
+
+        Parameters
+        ----------
+        value : str
+            Prompt template format.
+        """
+        return self._set(promptTemplate=value)
+
+    def setCustomPromptTemplate(self, value: str):
+        """Sets custom prompt template for image models.
+
+        Parameters
+        ----------
+        value : str
+            Custom prompt template string.
+        """
+        return self._set(customPromptTemplate=value)

sparknlp/reader/reader2table.py

@@ -0,0 +1,44 @@
+#  Copyright 2017-2025 John Snow Labs
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+from pyspark import keyword_only
+
+from sparknlp.common import AnnotatorType
+from sparknlp.internal import AnnotatorTransformer
+from sparknlp.partition.partition_properties import *
+
+
+class Reader2Table(
+    AnnotatorTransformer,
+    HasReaderProperties,
+    HasEmailReaderProperties,
+    HasExcelReaderProperties,
+    HasHTMLReaderProperties,
+    HasPowerPointProperties,
+    HasTextReaderProperties
+):
+    name = 'Reader2Table'
+
+    outputAnnotatorType = AnnotatorType.DOCUMENT
+
+    @keyword_only
+    def __init__(self):
+        super(Reader2Table, self).__init__(classname="com.johnsnowlabs.reader.Reader2Table")
+        self._setDefault(outputCol="document", outputFormat="json-table", inferTableStructure=True,
+                         outputAsDocument=False)
+
+    @keyword_only
+    def setParams(self):
+        kwargs = self._input_kwargs
+        return self._set(**kwargs)

sparknlp/reader/reader_assembler.py

@@ -0,0 +1,159 @@
+#  Copyright 2017-2025 John Snow Labs
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+
+from pyspark import keyword_only
+
+from sparknlp.common import AnnotatorType
+from sparknlp.internal import AnnotatorTransformer
+from sparknlp.partition.partition_properties import *
+
+class ReaderAssembler(
+    AnnotatorTransformer,
+    HasReaderProperties,
+    HasHTMLReaderProperties,
+    HasEmailReaderProperties,
+    HasExcelReaderProperties,
+    HasPowerPointProperties,
+    HasTextReaderProperties,
+    HasPdfProperties
+):
+    """
+    The ReaderAssembler annotator provides a unified interface for combining multiple Spark NLP
+    readers (such as Reader2Doc, Reader2Table, and Reader2Image) into a single, configurable
+    component. It automatically orchestrates the execution of different readers based on input type,
+    configured priorities, and fallback strategies allowing you to handle diverse content formats
+    without manually chaining multiple readers in your pipeline.
+
+    ReaderAssembler simplifies the process of building flexible pipelines capable of ingesting and
+    processing documents, tables, and images in a consistent way. It handles reader selection,
+    ordering, and fault-tolerance internally, ensuring that pipelines remain concise, robust, and
+    easy to maintain.
+
+    Examples
+    --------
+    >>> from johnsnowlabs.reader import ReaderAssembler
+    >>> from pyspark.ml import Pipeline
+    >>>
+    >>> reader_assembler = ReaderAssembler() \\
+    ...     .setContentType("text/html") \\
+    ...     .setContentPath("/table-image.html") \\
+    ...     .setOutputCol("document")
+    >>>
+    >>> pipeline = Pipeline(stages=[reader_assembler])
+    >>> pipeline_model = pipeline.fit(empty_data_set)
+    >>> result_df = pipeline_model.transform(empty_data_set)
+    >>>
+    >>> result_df.show()
+    +--------+--------------------+--------------------+--------------------+---------+
+    |fileName|       document_text|      document_table|      document_image|exception|
+    +--------+--------------------+--------------------+--------------------+---------+
+    |    null|[{'document', 0, 26...|[{'document', 0, 50...|[{'image', , 5, 5, ...|     null|
+    +--------+--------------------+--------------------+--------------------+---------+
+
+    This annotator is especially useful when working with heterogeneous input data — for example,
+    when a dataset includes PDFs, spreadsheets, and images — allowing Spark NLP to automatically
+    invoke the appropriate reader for each file type while preserving a unified schema in the output.
+"""
+
+
+    name = 'ReaderAssembler'
+
+    outputAnnotatorType = AnnotatorType.DOCUMENT
+
+    excludeNonText = Param(
+        Params._dummy(),
+        "excludeNonText",
+        "Whether to exclude non-text content from the output. Default is False.",
+        typeConverter=TypeConverters.toBoolean
+    )
+
+    userMessage = Param(
+        Params._dummy(),
+        "userMessage",
+        "Custom user message.",
+        typeConverter=TypeConverters.toString
+    )
+
+    promptTemplate = Param(
+        Params._dummy(),
+        "promptTemplate",
+        "Format of the output prompt.",
+        typeConverter=TypeConverters.toString
+    )
+
+    customPromptTemplate = Param(
+        Params._dummy(),
+        "customPromptTemplate",
+        "Custom prompt template for image models.",
+        typeConverter=TypeConverters.toString
+    )
+
+    @keyword_only
+    def __init__(self):
+        super(ReaderAssembler, self).__init__(classname="com.johnsnowlabs.reader.ReaderAssembler")
+        self._setDefault(contentType="",
+                         explodeDocs=False,
+                         userMessage="Describe this image",
+                         promptTemplate="qwen2vl-chat",
+                         readAsImage=True,
+                         customPromptTemplate="",
+                         ignoreExceptions=True,
+                         flattenOutput=False,
+                         titleThreshold=18)
+
+
+    @keyword_only
+    def setParams(self):
+        kwargs = self._input_kwargs
+        return self._set(**kwargs)
+
+    def setExcludeNonText(self, value):
+        """Sets whether to exclude non-text content from the output.
+
+        Parameters
+        ----------
+        value : bool
+            Whether to exclude non-text content from the output. Default is False.
+        """
+        return self._set(excludeNonText=value)
+
+    def setUserMessage(self, value: str):
+        """Sets custom user message.
+
+        Parameters
+        ----------
+        value : str
+            Custom user message to include.
+        """
+        return self._set(userMessage=value)
+
+    def setPromptTemplate(self, value: str):
+        """Sets format of the output prompt.
+
+        Parameters
+        ----------
+        value : str
+            Prompt template format.
+        """
+        return self._set(promptTemplate=value)
+
+    def setCustomPromptTemplate(self, value: str):
+        """Sets custom prompt template for image models.
+
+        Parameters
+        ----------
+        value : str
+            Custom prompt template string.
+        """
+        return self._set(customPromptTemplate=value)