- LaTeX table support

DHParser/ebnf.py

@@ -361,7 +361,7 @@ class EBNFCompiler(Compiler):
     WHITESPACE_KEYWORD = "WSP__"
     RESERVED_SYMBOLS = {WHITESPACE_KEYWORD, COMMENT_KEYWORD}
     AST_ERROR = "Badly structured syntax tree. " \
-                "Potentially due to erroneuos AST transformation."
+                "Potentially due to erroneous AST transformation."
     PREFIX_TABLE = {'§': 'Required',
                     '&': 'Lookahead', '!': 'NegativeLookahead',
                     '-&': 'Lookbehind', '-!': 'NegativeLookbehind',
@@ -591,7 +591,11 @@ class EBNFCompiler(Compiler):
     def on_definition(self, node: Node) -> Tuple[str, str]:
         rule = str(node.children[0])
         if rule in self.rules:
-            node.add_error('A rule with name "%s" has already been defined.' % rule)
+            first = self.rules[rule][0]
+            if not first._errors:
+                first.add_error('First definition of rule "%s" '
+                                'followed by illegal redefinitions.' % rule)
+            node.add_error('A rule with name "%s" has already been defined earlier.' % rule)
         elif rule in EBNFCompiler.RESERVED_SYMBOLS:
             node.add_error('Symbol "%s" is a reserved symbol.' % rule)
         elif not sane_parser_name(rule):

DHParser/syntaxtree.py

@@ -164,6 +164,8 @@ class Node:
             example by calling ``isinstance(node.parer, ...)``.
         errors (list):  A list of parser- or compiler-errors:
             tuple(position, string) attached to this node
+        error_flag (bool):  True, if either the node or any of its
+            descendants has errors.
         len (int):  The full length of the node's string result if the
             node is a leaf node or, otherwise, the concatenated string
             result's of its descendants. The figure always represents

DHParser/toolkit.py

@@ -193,7 +193,7 @@ def error_messages(source_text, errors) -> List[str]:
         a list that contains all error messages in string form. Each
         string starts with "line: [Line-No], column: [Column-No]
     """
-    return ["line: %i, column: %i" % line_col(source_text, err.pos) + ", error: %s" % err.msg
+    return ["line: %3i, column: %2i" % line_col(source_text, err.pos) + ", error: %s" % err.msg
             for err in sorted(list(errors))]
 
 

DHParser/transform.py

@@ -226,7 +226,7 @@ def traverse(root_node: Node,
     # # Code without optimization
     # table = {name: smart_list(call) for name, call in list(processing_table.items())}
     # table = expand_table(table)
-    # cache = {}
+    # cache = {}  # type: Dict[str, List[Callable]]
 
     def traverse_recursive(context):
         node = context[-1]
@@ -234,7 +234,7 @@ def traverse(root_node: Node,
             for child in node.result:
                 context.append(child)
                 traverse_recursive(context)  # depth first
-                node.error_flag |= child.error_flag  # propagate error flag
+                node.error_flag = node.error_flag or child.error_flag  # propagate error flag
                 context.pop()
 
         key = key_func(node)

examples/LaTeX/LaTeX.ebnf

@@ -68,14 +68,14 @@ figure              = "\begin{figure}" sequence §"\end{figure}"
 quotation           = ("\begin{quotation}" sequence §"\end{quotation}")
                     | ("\begin{quote}" sequence §"\end{quote}")
 verbatim            = "\begin{verbatim}" sequence §"\end{verbatim}"
-tabular             = "\begin{tabular}" tabular_config { tabular_cell } §"\end{tabular}"
-tabular_cell        = sequence
+tabular             = "\begin{tabular}" tabular_config { tabular_row } §"\end{tabular}"
+tabular_row         = [multicolumn | tabular_cell] { "&" [multicolumn | tabular_cell] } "\\" [ hline | cline ]
+tabular_cell        = { text_element //~ }+
 tabular_config      = "{" /[lcr|]+/~ §"}"
 
-
 #### paragraphs and sequences of paragraphs ####
 
-block_of_paragraphs = /{/~ sequence §/}/
+block_of_paragraphs = /{/~ [sequence] §/}/
 sequence            = { (paragraph | block_environment ) [PARSEP] }+
 paragraph           = { !blockcmd (text_element | LINEFEED) //~ }+
 text_element        = text | block | inline_environment | command
@@ -93,19 +93,20 @@ end_environment     = /\\end{/ §::NAME §/}/
 
 inline_math         = /\$/ /[^$]*/ §/\$/
 
-
 #### commands ####
 
 command             = known_command | text_command | generic_command
-known_command       = footnote | includegraphics | caption | multicolumn
+known_command       = footnote | includegraphics | caption | multicolumn | hline | cline
 text_command        = TXTCOMMAND | ESCAPED | BRACKETS
 generic_command     = !no_command CMDNAME [[ //~ config ] //~ block ]
 
 footnote            = "\footnote" block_of_paragraphs
 includegraphics     = "\includegraphics" [ config ] block
 caption             = "\caption" block
-multicolumn         = "\multicolumn" "{" INTEGER "}"
-                      tabular_config block_of_paragraphs
+multicolumn         = "\multicolumn" "{" INTEGER "}" tabular_config block_of_paragraphs
+hline               = "\hline"
+cline               = "\cline{" INTEGER "-" INTEGER "}"
+
 
 #######################################################################
 #
@@ -115,7 +116,7 @@ multicolumn         = "\multicolumn" "{" INTEGER "}"
 
 
 config     = "[" cfg_text §"]"
-cfg_text   = { ([//~] text) | CMDNAME }
+cfg_text   = { ([//~] text) | CMDNAME | SPECIAL }
 block      = /{/ //~ { !blockcmd text_element //~ } §/}/
 text       = TEXTCHUNK { //~ TEXTCHUNK }
 
@@ -139,6 +140,7 @@ structural = "subsection" | "section" | "chapter" | "subsubsection"
 CMDNAME    = /\\(?:(?!_)\w)+/~
 TXTCOMMAND = /\\text\w+/
 ESCAPED    = /\\[%$&_\/{}]/
+SPECIAL    = /[$&_\\\\\/]/
 BRACKETS   = /[\[\]]/                       # left or right square bracket: [ ]
 LINEFEED   = /[\\][\\]/
 

examples/LaTeX/LaTeXCompiler.py

@@ -117,14 +117,14 @@ class LaTeXGrammar(Grammar):
     quotation           = ("\begin{quotation}" sequence §"\end{quotation}")
                         | ("\begin{quote}" sequence §"\end{quote}")
     verbatim            = "\begin{verbatim}" sequence §"\end{verbatim}"
-    tabular             = "\begin{tabular}" tabular_config { tabular_cell } §"\end{tabular}"
-    tabular_cell        = sequence
+    tabular             = "\begin{tabular}" tabular_config { tabular_row } §"\end{tabular}"
+    tabular_row         = [multicolumn | tabular_cell] { "&" [multicolumn | tabular_cell] } "\\" [ hline | cline ]
+    tabular_cell        = { text_element //~ }+
     tabular_config      = "{" /[lcr|]+/~ §"}"
     
-    
     #### paragraphs and sequences of paragraphs ####
     
-    block_of_paragraphs = /{/~ sequence §/}/
+    block_of_paragraphs = /{/~ [sequence] §/}/
     sequence            = { (paragraph | block_environment ) [PARSEP] }+
     paragraph           = { !blockcmd (text_element | LINEFEED) //~ }+
     text_element        = text | block | inline_environment | command
@@ -142,19 +142,20 @@ class LaTeXGrammar(Grammar):
     
     inline_math         = /\$/ /[^$]*/ §/\$/
     
-    
     #### commands ####
     
     command             = known_command | text_command | generic_command
-    known_command       = footnote | includegraphics | caption | multicolumn
+    known_command       = footnote | includegraphics | caption | multicolumn | hline | cline
     text_command        = TXTCOMMAND | ESCAPED | BRACKETS
     generic_command     = !no_command CMDNAME [[ //~ config ] //~ block ]
     
     footnote            = "\footnote" block_of_paragraphs
     includegraphics     = "\includegraphics" [ config ] block
     caption             = "\caption" block
-    multicolumn         = "\multicolumn" "{" INTEGER "}"
-                          tabular_config block_of_paragraphs
+    multicolumn         = "\multicolumn" "{" INTEGER "}" tabular_config block_of_paragraphs
+    hline               = "\hline"
+    cline               = "\cline{" INTEGER "-" INTEGER "}"
+    
     
     #######################################################################
     #
@@ -164,7 +165,7 @@ class LaTeXGrammar(Grammar):
     
     
     config     = "[" cfg_text §"]"
-    cfg_text   = { ([//~] text) | CMDNAME }
+    cfg_text   = { ([//~] text) | CMDNAME | SPECIAL }
     block      = /{/ //~ { !blockcmd text_element //~ } §/}/
     text       = TEXTCHUNK { //~ TEXTCHUNK }
     
@@ -188,6 +189,7 @@ class LaTeXGrammar(Grammar):
     CMDNAME    = /\\(?:(?!_)\w)+/~
     TXTCOMMAND = /\\text\w+/
     ESCAPED    = /\\[%$&_\/{}]/
+    SPECIAL    = /[$&_\\\\\/]/
     BRACKETS   = /[\[\]]/                       # left or right square bracket: [ ]
     LINEFEED   = /[\\][\\]/
     
@@ -216,7 +218,7 @@ class LaTeXGrammar(Grammar):
     paragraph = Forward()
     tabular_config = Forward()
     text_element = Forward()
-    source_hash__ = "a99d24bcde48695ca003d544738b8057"
+    source_hash__ = "4003d206b4ecbd76dd8be99df87af4c0"
     parser_initialization__ = "upon instantiation"
     COMMENT__ = r'%.*(?:\n|$)'
     WSP__ = mixin_comment(whitespace=r'[ \t]*(?:\n(?![ \t]*\n)[ \t]*)?', comment=r'%.*(?:\n|$)')
@@ -235,6 +237,7 @@ class LaTeXGrammar(Grammar):
     NAME = Capture(RE('\\w+'))
     LINEFEED = RegExp('[\\\\][\\\\]')
     BRACKETS = RegExp('[\\[\\]]')
+    SPECIAL = RegExp('[$&_\\\\\\\\/]')
     ESCAPED = RegExp('\\\\[%$&_/{}]')
     TXTCOMMAND = RegExp('\\\\text\\w+')
     CMDNAME = RE('\\\\(?:(?!_)\\w)+')
@@ -243,15 +246,17 @@ class LaTeXGrammar(Grammar):
     no_command = Alternative(Token("\\begin{"), Token("\\end"), Series(BACKSLASH, structural))
     text = Series(TEXTCHUNK, ZeroOrMore(Series(RE(''), TEXTCHUNK)))
     block = Series(RegExp('{'), RE(''), ZeroOrMore(Series(NegativeLookahead(blockcmd), text_element, RE(''))), Required(RegExp('}')))
-    cfg_text = ZeroOrMore(Alternative(Series(Optional(RE('')), text), CMDNAME))
+    cfg_text = ZeroOrMore(Alternative(Series(Optional(RE('')), text), CMDNAME, SPECIAL))
     config = Series(Token("["), cfg_text, Required(Token("]")))
+    cline = Series(Token("\\cline{"), INTEGER, Token("-"), INTEGER, Token("}"))
+    hline = Token("\\hline")
     multicolumn = Series(Token("\\multicolumn"), Token("{"), INTEGER, Token("}"), tabular_config, block_of_paragraphs)
     caption = Series(Token("\\caption"), block)
     includegraphics = Series(Token("\\includegraphics"), Optional(config), block)
     footnote = Series(Token("\\footnote"), block_of_paragraphs)
     generic_command = Series(NegativeLookahead(no_command), CMDNAME, Optional(Series(Optional(Series(RE(''), config)), RE(''), block)))
     text_command = Alternative(TXTCOMMAND, ESCAPED, BRACKETS)
-    known_command = Alternative(footnote, includegraphics, caption, multicolumn)
+    known_command = Alternative(footnote, includegraphics, caption, multicolumn, hline, cline)
     command = Alternative(known_command, text_command, generic_command)
     inline_math = Series(RegExp('\\$'), RegExp('[^$]*'), Required(RegExp('\\$')))
     end_environment = Series(RegExp('\\\\end{'), Required(Pop(NAME)), Required(RegExp('}')))
@@ -262,13 +267,13 @@ class LaTeXGrammar(Grammar):
     known_inline_env = Synonym(inline_math)
     inline_environment = Alternative(known_inline_env, generic_inline_env)
     text_element.set(Alternative(text, block, inline_environment, command))
-    paragraph.set(
-        OneOrMore(Series(NegativeLookahead(blockcmd), Alternative(text_element, LINEFEED), RE(''))))
+    paragraph.set(OneOrMore(Series(NegativeLookahead(blockcmd), Alternative(text_element, LINEFEED), RE(''))))
     sequence = OneOrMore(Series(Alternative(paragraph, block_environment), Optional(PARSEP)))
-    block_of_paragraphs.set(Series(RE('{'), sequence, Required(RegExp('}'))))
+    block_of_paragraphs.set(Series(RE('{'), Optional(sequence), Required(RegExp('}'))))
     tabular_config.set(Series(Token("{"), RE('[lcr|]+'), Required(Token("}"))))
-    tabular_cell = Synonym(sequence)
-    tabular = Series(Token("\\begin{tabular}"), tabular_config, ZeroOrMore(tabular_cell), Required(Token("\\end{tabular}")))
+    tabular_cell = OneOrMore(Series(text_element, RE('')))
+    tabular_row = Series(Optional(Alternative(multicolumn, tabular_cell)), ZeroOrMore(Series(Token("&"), Optional(Alternative(multicolumn, tabular_cell)))), Token("\\\\"), Optional(Alternative(hline, cline)))
+    tabular = Series(Token("\\begin{tabular}"), tabular_config, ZeroOrMore(tabular_row), Required(Token("\\end{tabular}")))
     verbatim = Series(Token("\\begin{verbatim}"), sequence, Required(Token("\\end{verbatim}")))
     quotation = Alternative(Series(Token("\\begin{quotation}"), sequence, Required(Token("\\end{quotation}"))), Series(Token("\\begin{quote}"), sequence, Required(Token("\\end{quote}"))))
     figure = Series(Token("\\begin{figure}"), sequence, Required(Token("\\end{figure}")))
@@ -304,10 +309,10 @@ def get_grammar() -> LaTeXGrammar:
     global thread_local_LaTeX_grammar_singleton
     try:
         grammar = thread_local_LaTeX_grammar_singleton
-        return grammar
     except NameError:
         thread_local_LaTeX_grammar_singleton = LaTeXGrammar()
-        return thread_local_LaTeX_grammar_singleton
+        grammar = thread_local_LaTeX_grammar_singleton
+    return grammar
 
 
 #######################################################################

examples/LaTeX/grammar_tests/05_test_commands.ini

 [match:command]
 1 : \includegraphics[width=\textwidth]{Graph.eps}
 
+
+2 : \multicolumn{1}{c}{ }
+
+3 : \multicolumn{2}{c|}{material}
+
+4 : \multicolumn{2}{c}{$\underbrace{\hspace{7cm}}_{Simulations}$}
+

examples/LaTeX/grammar_tests/PENDING/test_tabular.ini → examples/LaTeX/grammar_tests/06_test_tabular.ini

@@ -3,7 +3,7 @@
               & $S_1$ \\ \cline{1-2}
         $A_1$ & $r_1$ \\ \cline{1-2}
         $A_2$ & $r_2$ \\ \cline{1-2}
-    \end{tabular}}
+    \end{tabular}
 
 2 : \begin{tabular}{c|c|c|c|cc|c|c|c|}
         \multicolumn{1}{c}{} & \multicolumn{3}{c}{Tabelle 1:} &