PyPDFForm 1.3.1__py3-none-any.whl → 1.3.3__py3-none-any.whl

PyPDFForm/__init__.py

@@ -5,4 +5,4 @@ from .wrapper import Wrapper
 
 PyPDFForm = Wrapper
 
-__version__ = "1.3.1"
+__version__ = "1.3.3"

PyPDFForm/core/constants.py

@@ -17,5 +17,12 @@ TEXT_FIELD_ALIGNMENT_IDENTIFIER = "/Q"
 CHOICE_FIELD_IDENTIFIER = "/Ch"
 CHOICES_IDENTIFIER = "/Opt"
 
+FONT_SIZE_IDENTIFIER = "Tf"
+FONT_COLOR_IDENTIFIER = " rg"
+DEFAULT_FONT = "Helvetica"
+DEFAULT_FONT_SIZE = 12
+DEFAULT_FONT_COLOR = (0, 0, 0)
+PREVIEW_FONT_COLOR = (1, 0, 0)
+
 CHECKBOX_TO_DRAW = "\u2713"
 RADIO_TO_DRAW = "\u25CF"

PyPDFForm/core/font.py

@@ -1,11 +1,17 @@
 # -*- coding: utf-8 -*-
 """Contains helpers for font."""
 
+import re
 from io import BytesIO
 
+import pdfrw
 from reportlab.pdfbase import pdfmetrics
 from reportlab.pdfbase.ttfonts import TTFError, TTFont
 
+from . import constants
+from .patterns import TEXT_FIELD_APPEARANCE_PATTERNS
+from .template import traverse_pattern
+
 
 def register_font(font_name: str, ttf_stream: bytes) -> bool:
     """Registers a font from a ttf file stream."""
@@ -22,3 +28,38 @@ def register_font(font_name: str, ttf_stream: bytes) -> bool:
 
     buff.close()
     return result
+
+
+def auto_detect_font(element: pdfrw.PdfDict) -> str:
+    """Returns the font of the text field if it is one of the standard fonts."""
+
+    result = constants.DEFAULT_FONT
+
+    for pattern in TEXT_FIELD_APPEARANCE_PATTERNS:
+        text_appearance = traverse_pattern(pattern, element)
+
+        if text_appearance:
+            text_appearance = (
+                text_appearance.replace("(", "").replace(")", "").split(" ")
+            )
+
+            for each in text_appearance:
+                if each.startswith("/"):
+                    text_segments = re.findall("[A-Z][^A-Z]*", each.replace("/", ""))
+
+                    for font in pdfmetrics.standardFonts:
+                        font_segments = re.findall(
+                            "[A-Z][^A-Z]*", font.replace("-", "")
+                        )
+                        if len(font_segments) != len(text_segments):
+                            continue
+
+                        found = True
+                        for i, val in enumerate(font_segments):
+                            if not val.startswith(text_segments[i]):
+                                found = False
+
+                        if found:
+                            return font
+
+    return result

PyPDFForm/core/font_size.py

@@ -6,8 +6,8 @@ from typing import Union
 
 import pdfrw
 
-from ..middleware.constants import GLOBAL_FONT_SIZE
 from . import constants, template
+from .constants import DEFAULT_FONT_SIZE
 
 
 def text_field_font_size(element: pdfrw.PdfDict) -> Union[float, int]:
@@ -17,7 +17,7 @@ def text_field_font_size(element: pdfrw.PdfDict) -> Union[float, int]:
     """
 
     if template.is_text_multiline(element):
-        return GLOBAL_FONT_SIZE
+        return DEFAULT_FONT_SIZE
 
     height = abs(
         float(element[constants.ANNOTATION_RECTANGLE_KEY][1])

PyPDFForm/core/template.py

@@ -148,13 +148,38 @@ def get_text_field_font_size(element: pdfrw.PdfDict) -> Union[float, int]:
     for pattern in TEXT_FIELD_APPEARANCE_PATTERNS:
         text_appearance = traverse_pattern(pattern, element)
         if text_appearance:
+            text_appearance = text_appearance.replace("(", "").replace(")", "")
             properties = text_appearance.split(" ")
-            if len(properties) > 1:
-                try:
-                    result = float(properties[1])
+            for i, val in enumerate(properties):
+                if val == constants.FONT_SIZE_IDENTIFIER:
+                    return float(properties[i - 1])
+
+    return result
+
+
+def get_text_field_font_color(
+    element: pdfrw.PdfDict,
+) -> Union[Tuple[float, float, float], None]:
+    """Returns the font color tuple of the text field if presented or black."""
+
+    result = (0, 0, 0)
+    for pattern in TEXT_FIELD_APPEARANCE_PATTERNS:
+        text_appearance = traverse_pattern(pattern, element)
+        if text_appearance:
+            if constants.FONT_COLOR_IDENTIFIER not in text_appearance:
+                return result
+
+            text_appearance = (
+                text_appearance.replace("(", "").replace(")", "").split(" ")
+            )
+            for i, val in enumerate(text_appearance):
+                if val == constants.FONT_COLOR_IDENTIFIER.replace(" ", ""):
+                    result = (
+                        float(text_appearance[i - 3]),
+                        float(text_appearance[i - 2]),
+                        float(text_appearance[i - 1]),
+                    )
                     break
-                except ValueError:
-                    pass
 
     return result
 
@@ -251,6 +276,12 @@ def get_draw_text_coordinates(
 ) -> Tuple[Union[float, int], Union[float, int]]:
     """Returns coordinates to draw text at given a PDF form text element."""
 
+    if element_middleware.preview:
+        return (
+            float(element[constants.ANNOTATION_RECTANGLE_KEY][0]),
+            float(element[constants.ANNOTATION_RECTANGLE_KEY][3]) + 5,
+        )
+
     element_value = element_middleware.value or ""
     length = (
         min(len(element_value), element_middleware.max_length)

PyPDFForm/core/utils.py

@@ -12,6 +12,7 @@ from ..middleware.constants import ELEMENT_TYPES
 from ..middleware.radio import Radio
 from ..middleware.text import Text
 from . import constants
+from . import font as font_core
 from . import font_size as font_size_core
 from . import template
 
@@ -43,10 +44,16 @@ def update_text_field_attributes(
             key = template.get_element_key(_element)
 
             if isinstance(elements[key], Text):
+                if elements[key].font is None:
+                    elements[key].font = font_core.auto_detect_font(_element)
                 if elements[key].font_size is None:
                     elements[key].font_size = template.get_text_field_font_size(
                         _element
                     ) or font_size_core.text_field_font_size(_element)
+                if elements[key].font_color is None:
+                    elements[key].font_color = template.get_text_field_font_color(
+                        _element
+                    )
                 if (
                     template.is_text_multiline(_element)
                     and elements[key].text_wrap_length is None
@@ -135,9 +142,9 @@ def checkbox_radio_to_draw(
         element_name=element.name,
         element_value="",
     )
-    new_element.font = "Helvetica"
+    new_element.font = constants.DEFAULT_FONT
     new_element.font_size = font_size
-    new_element.font_color = (0, 0, 0)
+    new_element.font_color = constants.DEFAULT_FONT_COLOR
 
     if isinstance(element, Checkbox):
         new_element.value = constants.CHECKBOX_TO_DRAW
@@ -147,6 +154,21 @@ def checkbox_radio_to_draw(
     return new_element
 
 
+def preview_element_to_draw(element: ELEMENT_TYPES) -> Text:
+    """Converts an element to a preview text element."""
+
+    new_element = Text(
+        element_name=element.name,
+        element_value="{" + f" {element.name} " + "}",
+    )
+    new_element.font = constants.DEFAULT_FONT
+    new_element.font_size = constants.DEFAULT_FONT_SIZE
+    new_element.font_color = constants.PREVIEW_FONT_COLOR
+    new_element.preview = True
+
+    return new_element
+
+
 def remove_all_elements(pdf: bytes) -> bytes:
     """Removes all elements from a pdfrw parsed PDF form."""
 

PyPDFForm/middleware/checkbox.py

@@ -21,3 +21,9 @@ class Checkbox(Element):
         """Json schema definition of the checkbox."""
 
         return {"type": "boolean"}
+
+    @property
+    def sample_value(self) -> bool:
+        """Sample value of the checkbox."""
+
+        return True

PyPDFForm/middleware/constants.py

@@ -8,10 +8,6 @@ from .dropdown import Dropdown
 from .radio import Radio
 from .text import Text
 
-GLOBAL_FONT = "Helvetica"
-GLOBAL_FONT_SIZE = 12
-GLOBAL_FONT_COLOR = (0, 0, 0)
-
 VERSION_IDENTIFIERS = [
     b"%PDF-1.0",
     b"%PDF-1.1",

PyPDFForm/middleware/dropdown.py

@@ -23,3 +23,9 @@ class Dropdown(Element):
         """Json schema definition of the dropdown."""
 
         return {"type": "integer", "maximum": len(self.choices) - 1}
+
+    @property
+    def sample_value(self) -> int:
+        """Sample value of the dropdown."""
+
+        return 0

PyPDFForm/middleware/element.py

@@ -1,7 +1,7 @@
 # -*- coding: utf-8 -*-
 """Contains element middleware."""
 
-from typing import Union
+from typing import Any, Union
 
 
 class Element:
@@ -28,3 +28,9 @@ class Element:
         """Json schema definition of the element."""
 
         raise NotImplementedError
+
+    @property
+    def sample_value(self) -> Any:
+        """Sample value of the element."""
+
+        raise NotImplementedError

PyPDFForm/middleware/radio.py

@@ -23,3 +23,9 @@ class Radio(Element):
         """Json schema definition of the radiobutton."""
 
         return {"type": "integer", "maximum": self.number_of_options - 1}
+
+    @property
+    def sample_value(self) -> int:
+        """Sample value of the radiobutton."""
+
+        return 0

PyPDFForm/middleware/template.py

@@ -67,9 +67,6 @@ def dropdown_to_text(dropdown: Dropdown) -> Text:
 
     result = Text(dropdown.name)
 
-    result.font = constants.GLOBAL_FONT
-    result.font_color = constants.GLOBAL_FONT_COLOR
-
     if dropdown.value is not None:
         result.value = (
             dropdown.choices[dropdown.value]

PyPDFForm/middleware/text.py

@@ -25,6 +25,7 @@ class Text(Element):
         self.character_paddings = None
         self.text_lines = None
         self.text_line_x_coordinates = None
+        self.preview = False
 
     @property
     def schema_definition(self) -> dict:
@@ -36,3 +37,9 @@ class Text(Element):
             result["maxLength"] = self.max_length
 
         return result
+
+    @property
+    def sample_value(self) -> str:
+        """Sample value of the text field."""
+
+        return self.name

PyPDFForm/wrapper.py

@@ -5,6 +5,7 @@ from __future__ import annotations
 
 from typing import BinaryIO, Dict, Union
 
+from .core import constants as core_constants
 from .core import filler, font
 from .core import image as image_core
 from .core import utils
@@ -32,17 +33,21 @@ class Wrapper:
 
         for each in self.elements.values():
             if isinstance(each, Text):
-                each.font = kwargs.get("global_font", constants.GLOBAL_FONT)
+                each.font = kwargs.get("global_font")
                 each.font_size = kwargs.get("global_font_size")
-                each.font_color = kwargs.get(
-                    "global_font_color", constants.GLOBAL_FONT_COLOR
-                )
+                each.font_color = kwargs.get("global_font_color")
 
     def read(self) -> bytes:
         """Reads the file stream of a PDF form."""
 
         return self.stream
 
+    @property
+    def sample_data(self) -> dict:
+        """Returns a valid sample data that can be filled into the PDF form."""
+
+        return {key: value.sample_value for key, value in self.elements.items()}
+
     @property
     def version(self) -> Union[str, None]:
         """Gets the version of the PDF."""
@@ -78,11 +83,23 @@ class Wrapper:
 
         return new_obj
 
+    @property
+    def preview(self) -> bytes:
+        """Inspects all supported elements' names for the PDF form."""
+
+        return filler.fill(
+            self.stream,
+            {
+                key: utils.preview_element_to_draw(value)
+                for key, value in self.elements.items()
+            },
+        )
+
     def fill(
         self,
         data: Dict[str, Union[str, bool, int]],
     ) -> Wrapper:
-        """Fill a PDF form."""
+        """Fills a PDF form."""
 
         for key, value in data.items():
             if key in self.elements:
@@ -114,9 +131,13 @@ class Wrapper:
 
         new_element = Text("new")
         new_element.value = text
-        new_element.font = kwargs.get("font", constants.GLOBAL_FONT)
-        new_element.font_size = kwargs.get("font_size", constants.GLOBAL_FONT_SIZE)
-        new_element.font_color = kwargs.get("font_color", constants.GLOBAL_FONT_COLOR)
+        new_element.font = kwargs.get("font", core_constants.DEFAULT_FONT)
+        new_element.font_size = kwargs.get(
+            "font_size", core_constants.DEFAULT_FONT_SIZE
+        )
+        new_element.font_color = kwargs.get(
+            "font_color", core_constants.DEFAULT_FONT_COLOR
+        )
 
         watermarks = watermark_core.create_watermarks_and_draw(
             self.stream,

{PyPDFForm-1.3.1.dist-info → PyPDFForm-1.3.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: PyPDFForm
-Version: 1.3.1
+Version: 1.3.3
 Summary: The Python library for PDF forms.
 Home-page: https://github.com/chinapandaman/PyPDFForm
 Author: Jinge Li
@@ -84,10 +84,6 @@ and it should look like [this](https://github.com/chinapandaman/PyPDFForm/raw/ma
 
 ## How to Contribute
 
-If you wish to improve this library, there is one specific way you can contribute 
-on top of the usual open source project norms such as issues and pull requests.
-
 It is difficult to make sure that the library supports all the PDF form creating tools out 
-there. So if you run into a case where the library does not work for certain PDF forms created by 
-certain tools, feel free to open an issue with the problematic PDF form attached. I will seek 
+there. So if you run into a case where the library does not work for certain PDF forms created by certain tools, feel free to open an issue with the problematic PDF form attached. I will seek 
 to make the library support the attached PDF form as well as the tool used to create it.

PyPDFForm-1.3.3.dist-info/RECORD

@@ -0,0 +1,26 @@
+PyPDFForm/__init__.py,sha256=ecXWLEcE7Yxn6rn5RQn17cDB5V7JsbhX1zFWAPFPfJs,122
+PyPDFForm/wrapper.py,sha256=jBAzndmoIX1-dHr7hnnIBMpuGqNNKQvC40yd2mdgyVI,6052
+PyPDFForm/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+PyPDFForm/core/constants.py,sha256=6q06MVJ_-u60MN3F7pp_uYDIZY1V7GQi7WR5PaX8Q5Q,752
+PyPDFForm/core/filler.py,sha256=XK_g6hHEMHUt99S0Fj_M85x_ChFptlYfYxvUeuGpviQ,2701
+PyPDFForm/core/font.py,sha256=RLiIC7nv5XaIC92fMfVcozrocCFFJ_IQWboLoNRG60E,1909
+PyPDFForm/core/font_size.py,sha256=Ussv1Ah-4qB-5PryJf7hy2UFszllJWlVP7iEYTfmvqQ,1153
+PyPDFForm/core/image.py,sha256=pwsGMGjvtmxRInF6H-LgGMJosJnZ0uPKS1LU59aeKms,1070
+PyPDFForm/core/patterns.py,sha256=AzYvpNppIGOBUXzZw2LCS8XBJ8F8GOmlBTFBY_OQej4,2402
+PyPDFForm/core/template.py,sha256=7SXXQ5_E4HzGCc3J0AikzsB20lEjgJkaAoMRFbEZN8E,12701
+PyPDFForm/core/utils.py,sha256=sMs_rxhrq4N3uTce5bPyKu2wEr-5hwvTDz-EFQ-sNBE,6062
+PyPDFForm/core/watermark.py,sha256=WAl1idxWbeMwHA1AxdsbLHEPCbniSHD4gGFXf4zlHl8,4242
+PyPDFForm/middleware/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+PyPDFForm/middleware/adapter.py,sha256=jen3jzgik_hwrM7lG1GdVi0CSzELaePv9UzRb0VIRPU,876
+PyPDFForm/middleware/checkbox.py,sha256=fU_VV6B6m7aYe3i6HzXBUu9U9yH5gP0WXz1J9q4c0Fs,654
+PyPDFForm/middleware/constants.py,sha256=r56j2oURQei7KX0gWPqb10bQXRVdyVd0T8BapSlDlYM,481
+PyPDFForm/middleware/dropdown.py,sha256=OJQI_XffMPf4yxUCk5WOaLIqZpRd97gNV-YG-llFaa8,712
+PyPDFForm/middleware/element.py,sha256=Xo_lFRCTkEorfYOSU7RcP7tmvCrkKEpiZAzKIPHRwVs,787
+PyPDFForm/middleware/radio.py,sha256=gwNTE__eiaSqxY8_ov__tI8byz-cKOfTp9nWJ5DkFzQ,730
+PyPDFForm/middleware/template.py,sha256=Dv0I70DQIsejnVZlrrebY8W8YHsMLDd8kxUIeXKkdGM,2354
+PyPDFForm/middleware/text.py,sha256=rrTfXcxaXZlLsjuJ1C4KaHt_reZk2c9MQ-z0ec3OD9g,1095
+PyPDFForm-1.3.3.dist-info/LICENSE,sha256=43awmYkI6opyTpg19me731iO1WfXZwViqb67oWtCsFY,1065
+PyPDFForm-1.3.3.dist-info/METADATA,sha256=2D-gpDc8csDOBEAaO_62857YVXYwtVbWysm9lzoScIg,3985
+PyPDFForm-1.3.3.dist-info/WHEEL,sha256=yQN5g4mg4AybRjkgi-9yy4iQEFibGQmlz78Pik5Or-A,92
+PyPDFForm-1.3.3.dist-info/top_level.txt,sha256=GQQKuWqPUjT9YZqwK95NlAQzxjwoQrsxQ8ureM8lWOY,10
+PyPDFForm-1.3.3.dist-info/RECORD,,

PyPDFForm-1.3.1.dist-info/RECORD

@@ -1,26 +0,0 @@
-PyPDFForm/__init__.py,sha256=VHAFcKg3dA0WhfJyqKvVEwdcD6155-RvfJFU2lXc5FQ,122
-PyPDFForm/wrapper.py,sha256=SANVsyJSMHMEH8sbiUgAI4t8helhKajRaNkq8WAYlKM,5496
-PyPDFForm/core/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-PyPDFForm/core/constants.py,sha256=uyzfuEpTNq1nwOI5ahjWKDEqtM7enEnMIY0TATs9toQ,581
-PyPDFForm/core/filler.py,sha256=XK_g6hHEMHUt99S0Fj_M85x_ChFptlYfYxvUeuGpviQ,2701
-PyPDFForm/core/font.py,sha256=9hbKyyz4QE1nwadUTtJgEBetfFW-8qwGLVaVDDOqbMI,542
-PyPDFForm/core/font_size.py,sha256=waJqMC2SHnexYfEEBhpDeapzBcpddIYp6tNkMaHH36Q,1163
-PyPDFForm/core/image.py,sha256=pwsGMGjvtmxRInF6H-LgGMJosJnZ0uPKS1LU59aeKms,1070
-PyPDFForm/core/patterns.py,sha256=AzYvpNppIGOBUXzZw2LCS8XBJ8F8GOmlBTFBY_OQej4,2402
-PyPDFForm/core/template.py,sha256=H2qQLJuwhr2zLtCq1CjtvxKaOfVqpLkYJmYsQkz2xFk,11485
-PyPDFForm/core/utils.py,sha256=3Ri3J09k9awUKUiiYjW1otJfhK1VPSEO2KfED7MrMaY,5228
-PyPDFForm/core/watermark.py,sha256=WAl1idxWbeMwHA1AxdsbLHEPCbniSHD4gGFXf4zlHl8,4242
-PyPDFForm/middleware/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-PyPDFForm/middleware/adapter.py,sha256=jen3jzgik_hwrM7lG1GdVi0CSzELaePv9UzRb0VIRPU,876
-PyPDFForm/middleware/checkbox.py,sha256=DX4yTNP5u-jRoe5GdGbVwfIGBXc84J5H0bvKW7VIdJk,538
-PyPDFForm/middleware/constants.py,sha256=qGFhOuToiscSFpPoVXnpfRkuo-flPpFKRa4yExAu5ZI,560
-PyPDFForm/middleware/dropdown.py,sha256=ZlS6X0Fq_Tqox3DuxGk4qQujrF1wXDFkWdYpMmQ2Nco,600
-PyPDFForm/middleware/element.py,sha256=prjp8rg_F4DRnHRsSIki0yLMw3OHZuSIgkbocVSlPKE,654
-PyPDFForm/middleware/radio.py,sha256=-1RiAmzui6yPUgUYowRw_gaEN1e2XPFgqfbF9GVanrI,615
-PyPDFForm/middleware/template.py,sha256=7-ttGCiwlY_P5h4sr660BLUkP-UVqv8K0f62NLRmIaw,2447
-PyPDFForm/middleware/text.py,sha256=jGW9M4QwWuSUBvOUgs2BoSYzp91F0tUg3_2uNmXPevY,944
-PyPDFForm-1.3.1.dist-info/LICENSE,sha256=43awmYkI6opyTpg19me731iO1WfXZwViqb67oWtCsFY,1065
-PyPDFForm-1.3.1.dist-info/METADATA,sha256=W_8srJBKPvYPP-1P-QgwGvYsfq8YC1mGvc82X5KBhfQ,4150
-PyPDFForm-1.3.1.dist-info/WHEEL,sha256=yQN5g4mg4AybRjkgi-9yy4iQEFibGQmlz78Pik5Or-A,92
-PyPDFForm-1.3.1.dist-info/top_level.txt,sha256=GQQKuWqPUjT9YZqwK95NlAQzxjwoQrsxQ8ureM8lWOY,10
-PyPDFForm-1.3.1.dist-info/RECORD,,