Skip to content

Adobe Loader

AdobeReader ΒΆ

Bases: BaseReader

Read PDF using the Adobe's PDF Services. Be able to extract text, table, and figure with high accuracy

Example 
1
2
3
>> from kotaemon.loaders import AdobeReader
>> reader = AdobeReader()
>> documents = reader.load_data("path/to/pdf")

Args: endpoint: URL to the Vision Language Model endpoint. If not provided, will use the default kotaemon.loaders.adobe_loader.DEFAULT_VLM_ENDPOINT

1
2
max_figures_to_caption: an int decides how many figured will be captioned.
The rest will be ignored (are indexed without captions).
Source code in libs/kotaemon/kotaemon/loaders/adobe_loader.py 
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
class AdobeReader(BaseReader):
    """Read PDF using the Adobe's PDF Services.
    Be able to extract text, table, and figure with high accuracy

    Example:
        ```python
        >> from kotaemon.loaders import AdobeReader
        >> reader = AdobeReader()
        >> documents = reader.load_data("path/to/pdf")
        ```
    Args:
        endpoint: URL to the Vision Language Model endpoint. If not provided,
        will use the default `kotaemon.loaders.adobe_loader.DEFAULT_VLM_ENDPOINT`

        max_figures_to_caption: an int decides how many figured will be captioned.
        The rest will be ignored (are indexed without captions).
    """

    def __init__(
        self,
        vlm_endpoint: Optional[str] = None,
        max_figures_to_caption: int = 100,
        *args: Any,
        **kwargs: Any,
    ) -> None:
        """Init params"""
        super().__init__(*args)
        self.table_regex = r"/Table(\[\d+\])?$"
        self.figure_regex = r"/Figure(\[\d+\])?$"
        self.vlm_endpoint = vlm_endpoint or DEFAULT_VLM_ENDPOINT
        self.max_figures_to_caption = max_figures_to_caption

    def load_data(
        self, file: Path, extra_info: Optional[Dict] = None, **kwargs
    ) -> List[Document]:
        """Load data by calling to the Adobe's API

        Args:
            file (Path): Path to the PDF file

        Returns:
            List[Document]: list of documents extracted from the PDF file,
                includes 3 types: text, table, and image

        """
        from .utils.adobe import (
            generate_figure_captions,
            load_json,
            parse_figure_paths,
            parse_table_paths,
            request_adobe_service,
        )

        filename = file.name
        filepath = str(Path(file).resolve())
        output_path = request_adobe_service(file_path=str(file), output_path="")
        results_path = os.path.join(output_path, "structuredData.json")

        if not os.path.exists(results_path):
            logger.exception("Fail to parse the document.")
            return []

        data = load_json(results_path)

        texts = defaultdict(list)
        tables = []
        figures = []

        elements = data["elements"]
        for item_id, item in enumerate(elements):
            page_number = item.get("Page", -1) + 1
            item_path = item["Path"]
            item_text = item.get("Text", "")

            file_paths = [
                Path(output_path) / path for path in item.get("filePaths", [])
            ]
            prev_item = elements[item_id - 1]
            title = prev_item.get("Text", "")

            if re.search(self.table_regex, item_path):
                table_content = parse_table_paths(file_paths)
                if not table_content:
                    continue
                table_caption = (
                    table_content.replace("|", "").replace("---", "")
                    + f"\n(Table in Page {page_number}. {title})"
                )
                tables.append((page_number, table_content, table_caption))

            elif re.search(self.figure_regex, item_path):
                figure_caption = (
                    item_text + f"\n(Figure in Page {page_number}. {title})"
                )
                figure_content = parse_figure_paths(file_paths)
                if not figure_content:
                    continue
                figures.append([page_number, figure_content, figure_caption])

            else:
                if item_text and "Table" not in item_path and "Figure" not in item_path:
                    texts[page_number].append(item_text)

        # get figure caption using GPT-4V
        figure_captions = generate_figure_captions(
            self.vlm_endpoint,
            [item[1] for item in figures],
            self.max_figures_to_caption,
        )
        for item, caption in zip(figures, figure_captions):
            # update figure caption
            item[2] += " " + caption

        # Wrap elements with Document
        documents = []

        # join plain text elements
        for page_number, txts in texts.items():
            documents.append(
                Document(
                    text="\n".join(txts),
                    metadata={
                        "page_label": page_number,
                        "file_name": filename,
                        "file_path": filepath,
                        **(extra_info if extra_info else {}),
                    },
                )
            )

        # table elements
        for page_number, table_content, table_caption in tables:
            documents.append(
                Document(
                    text=table_content,
                    metadata={
                        "table_origin": table_content,
                        "type": "table",
                        "page_label": page_number,
                        "file_name": filename,
                        "file_path": filepath,
                        **(extra_info if extra_info else {}),
                    },
                    metadata_template="",
                    metadata_seperator="",
                )
            )

        # figure elements
        for page_number, figure_content, figure_caption in figures:
            documents.append(
                Document(
                    text=figure_caption,
                    metadata={
                        "image_origin": figure_content,
                        "type": "image",
                        "page_label": page_number,
                        "file_name": filename,
                        "file_path": filepath,
                        **(extra_info if extra_info else {}),
                    },
                    metadata_template="",
                    metadata_seperator="",
                )
            )
        return documents

load_data ΒΆ

1
load_data(file, extra_info=None, **kwargs)

Load data by calling to the Adobe's API

Parameters:

Name Type Description Default
file Path

Path to the PDF file

 required

Returns:

Type Description
List[Document]

List[Document]: list of documents extracted from the PDF file, includes 3 types: text, table, and image
Source code in libs/kotaemon/kotaemon/loaders/adobe_loader.py 
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
def load_data(
    self, file: Path, extra_info: Optional[Dict] = None, **kwargs
) -> List[Document]:
    """Load data by calling to the Adobe's API

    Args:
        file (Path): Path to the PDF file

    Returns:
        List[Document]: list of documents extracted from the PDF file,
            includes 3 types: text, table, and image

    """
    from .utils.adobe import (
        generate_figure_captions,
        load_json,
        parse_figure_paths,
        parse_table_paths,
        request_adobe_service,
    )

    filename = file.name
    filepath = str(Path(file).resolve())
    output_path = request_adobe_service(file_path=str(file), output_path="")
    results_path = os.path.join(output_path, "structuredData.json")

    if not os.path.exists(results_path):
        logger.exception("Fail to parse the document.")
        return []

    data = load_json(results_path)

    texts = defaultdict(list)
    tables = []
    figures = []

    elements = data["elements"]
    for item_id, item in enumerate(elements):
        page_number = item.get("Page", -1) + 1
        item_path = item["Path"]
        item_text = item.get("Text", "")

        file_paths = [
            Path(output_path) / path for path in item.get("filePaths", [])
        ]
        prev_item = elements[item_id - 1]
        title = prev_item.get("Text", "")

        if re.search(self.table_regex, item_path):
            table_content = parse_table_paths(file_paths)
            if not table_content:
                continue
            table_caption = (
                table_content.replace("|", "").replace("---", "")
                + f"\n(Table in Page {page_number}. {title})"
            )
            tables.append((page_number, table_content, table_caption))

        elif re.search(self.figure_regex, item_path):
            figure_caption = (
                item_text + f"\n(Figure in Page {page_number}. {title})"
            )
            figure_content = parse_figure_paths(file_paths)
            if not figure_content:
                continue
            figures.append([page_number, figure_content, figure_caption])

        else:
            if item_text and "Table" not in item_path and "Figure" not in item_path:
                texts[page_number].append(item_text)

    # get figure caption using GPT-4V
    figure_captions = generate_figure_captions(
        self.vlm_endpoint,
        [item[1] for item in figures],
        self.max_figures_to_caption,
    )
    for item, caption in zip(figures, figure_captions):
        # update figure caption
        item[2] += " " + caption

    # Wrap elements with Document
    documents = []

    # join plain text elements
    for page_number, txts in texts.items():
        documents.append(
            Document(
                text="\n".join(txts),
                metadata={
                    "page_label": page_number,
                    "file_name": filename,
                    "file_path": filepath,
                    **(extra_info if extra_info else {}),
                },
            )
        )

    # table elements
    for page_number, table_content, table_caption in tables:
        documents.append(
            Document(
                text=table_content,
                metadata={
                    "table_origin": table_content,
                    "type": "table",
                    "page_label": page_number,
                    "file_name": filename,
                    "file_path": filepath,
                    **(extra_info if extra_info else {}),
                },
                metadata_template="",
                metadata_seperator="",
            )
        )

    # figure elements
    for page_number, figure_content, figure_caption in figures:
        documents.append(
            Document(
                text=figure_caption,
                metadata={
                    "image_origin": figure_content,
                    "type": "image",
                    "page_label": page_number,
                    "file_name": filename,
                    "file_path": filepath,
                    **(extra_info if extra_info else {}),
                },
                metadata_template="",
                metadata_seperator="",
            )
        )
    return documents