rtfparser can now handle \binN control words

Author: Sven Siegmund

src/rtfparse/entities.py

@@ -69,6 +69,7 @@ def __init__(self, config: config_loader.Config, file: io.BufferedReader) -> Non
         logger.debug(f"Reading Control Word at file position {file.tell()}")
         self.control_name = "missing"
         self.parameter = ""
+        self.bindata = b""
         self.start_position = file.tell()
         logger.debug(f"Starting at file position {self.start_position}")
         probe = file.read(CONTROL_WORD)
@@ -86,6 +87,9 @@ def __init__(self, config: config_loader.Config, file: io.BufferedReader) -> Non
                 logger.debug(f"Delimiter is {match.group('other').decode(self.config.default_encoding)}, len: {len(match.group('delimiter'))}")
                 target_position -= len(match.group("delimiter"))
             file.seek(target_position)
+            # handle \binN:
+            if self.control_name == "bin":
+               self.bindata = file.read(utils.twos_complement(self.parameter, INTEGER_MAGNITUDE))
         else:
             logger.warning(f"Missing Control Word")
             file.seek(self.start_position)

src/rtfparse/entry.py

@@ -82,7 +82,6 @@ def run(config: config_loader.Config) -> None:
         de_encapsulate(rp, config.cli_args.de_encapsulate_html)
 
 
-
 def cli_start(version) -> None:
     """
     Entry point for any component start from the commmand line

src/rtfparse/minimal.py

@@ -0,0 +1,17 @@
+#!/usr/bin/env python
+
+
+def tc(val, nbits):
+    """Compute the 2's complement of int value val. Credit: https://stackoverflow.com/a/37075643/9235421"""
+    if val < 0:
+        if (val + 1).bit_length() >= nbits:
+            raise ValueError(f"Value {val} is out of range of {nbits}-bit value.")
+        val = (1 << nbits) + val
+    else:
+        if val.bit_length() > nbits:
+            raise ValueError(f"Value {val} is out of range of {nbits}-bit value.")
+        # If sign bit is set.
+        if (val & (1 << (nbits - 1))) != 0:
+            # compute negative value.
+            val = val - (1 << nbits)
+    return val

src/rtfparse/re_patterns.py

@@ -56,11 +56,12 @@ def no_capture(content: bytes) -> bytes:
 # int16 = minus + digit + rb"{1,5}"
 parameter_pattern = named_regex_group("parameter", digit)
 space = named_regex_group("space", rb" ")
+newline = named_regex_group("newline", _newline)
 other = named_regex_group("other", group(rb"^" + _letters + _digits))
 
 
 ascii_letter_sequence = named_regex_group("control_name", ascii_letters + parameter_pattern + rb"?")
-delimiter = named_regex_group("delimiter", rb"|".join((space, other, rb"$")))
+delimiter = named_regex_group("delimiter", rb"|".join((space, newline, other, rb"$")))
 symbol = named_regex_group("symbol", other)
 control_word_pattern = named_regex_group("control_word", rtf_backslash + ascii_letter_sequence + delimiter)
 pcdata_delimiter = no_capture(rb"|".join((rtf_brace_open, rtf_brace_close, control_word_pattern)))

src/rtfparse/utils.py

@@ -51,3 +51,19 @@ def what_is_being_parsed(file: Union[io.BufferedReader, io.BytesIO]) -> str:
         return file.name
     elif isinstance(file, io.BytesIO):
         return repr(file)
+
+
+def twos_complement(val, nbits):
+    """Compute the 2's complement of int value val. Credit: https://stackoverflow.com/a/37075643/9235421"""
+    if val < 0:
+        if (val + 1).bit_length() >= nbits:
+            raise ValueError(f"Value {val} is out of range of {nbits}-bit value.")
+        val = (1 << nbits) + val
+    else:
+        if val.bit_length() > nbits:
+            raise ValueError(f"Value {val} is out of range of {nbits}-bit value.")
+        # If sign bit is set.
+        if (val & (1 << (nbits - 1))) != 0:
+            # compute negative value.
+            val = val - (1 << nbits)
+    return val