" or "<{Component} title=test/>".
+
+ Or more often "<{Component} title={title}/>" which should be corrected
+ with "<{Component} title={title} />".
+ """
+ if sinfo is not None:
+ return (
+ # has attributes
+ len(sinfo.raw_attrs) > 0
+ # last attr not bare attribute
+ and sinfo.raw_attrs[-1][1] is not None
+ # last char of last attr is "/"
+ and sinfo.raw_attrs[-1][1][-1] == "/"
+ # parsed starttag ends with "/>"
+ and sinfo.starttag_text.endswith("/>")
+ # if parsed as startend then its not ambiguous
+ and not sinfo.startend
+ )
+ return False
+
# ------------------------------------------
# HTMLParser tag callbacks
# ------------------------------------------
@@ -385,19 +531,60 @@ def handle_starttag(self, tag: str, attrs: Sequence[HTMLAttribute]) -> None:
def handle_startendtag(self, tag: str, attrs: Sequence[HTMLAttribute]) -> None:
"""Dispatch a self-closing tag, `
` to specialized handlers."""
- open_tag = self.make_open_tag(tag, attrs)
+ open_tag = self.make_open_tag(tag, attrs, startend=True)
final_tag = self.finalize_tag(open_tag)
self.append_child(final_tag)
def handle_endtag(self, tag: str) -> None:
if not self.stack:
- raise ValueError(f"Unexpected closing tag with no open tag.")
-
+ tag_ref = self.placeholders.copy().remove_placeholders(tag)
+ if tag_ref.is_literal:
+ raise ValueError(f"Unexpected closing tag with no open tag.")
+ if not tag_ref.is_singleton:
+ # @TODO: Also it doesn't match anything
+ raise ValueError(
+ "Component end tags must have exactly one interpolation."
+ )
+ # Component tag endtag but no component tag is open...
+ unmatched_endtag = self.get_source().format_endtag(tag_ref.i_indexes[0])
+ raise ValueError(
+ f"Unexpected closing component tag with no open tag."
+ )
open_tag = self.stack.pop()
endtag_i_index = self.validate_end_tag(tag, open_tag)
- final_tag = self.finalize_tag(open_tag, endtag_i_index)
+ final_tag = self.finalize_tag(
+ open_tag, endtag_i_index=endtag_i_index, endtag_pos=self.get_parser_pos()
+ )
self.append_child(final_tag)
+ def get_closed_tcomps(
+ self, root: OpenTag | None, recurse_component_children: bool = False
+ ) -> list[TComponent]:
+ """
+ Get TComponents that were closed during parsing starting from `root`.
+
+ If `root` is None then use the parser's default `root`.
+
+ TComponents should be returned in the order they were closed in:
+ from first closed to last closed.
+
+ @NOTE: That the root is an `OpenTag` but its `children` are actually `TNode`s.
+ """
+ if root is None:
+ root = self.root
+ tcomps = []
+ nodes = list(root.children)
+ while nodes:
+ node = nodes.pop()
+ if isinstance(node, TComponent):
+ tcomps.append(node)
+ if recurse_component_children:
+ children = self.tcomponent_children.get(node, [])
+ nodes.extend(children)
+ elif isinstance(node, (TElement, TFragment)):
+ nodes.extend(node.children)
+ return tcomps
+
# ------------------------------------------
# HTMLParser other callbacks
# ------------------------------------------
@@ -406,15 +593,18 @@ def handle_data(self, data: str) -> None:
ref = self.placeholders.remove_placeholders(data)
parent = self.get_parent()
if parent.children and isinstance(parent.children[-1], TText):
+ prior_text = parent.children[-1]
parent.children[-1] = TText(
- ref=combine_template_refs(parent.children[-1].ref, ref)
+ ref=combine_template_refs(prior_text.ref, ref),
+ # Keep starting position of the prior text
+ parser_pos=prior_text.parser_pos,
)
else:
- self.append_child(TText(ref=ref))
+ self.append_child(TText(ref=ref, parser_pos=self.get_parser_pos()))
def handle_comment(self, data: str) -> None:
ref = self.placeholders.remove_placeholders(data)
- comment = TComment(ref)
+ comment = TComment(ref, parser_pos=self.get_parser_pos())
self.append_child(comment)
def handle_decl(self, decl: str) -> None:
@@ -423,7 +613,7 @@ def handle_decl(self, decl: str) -> None:
raise ValueError("Interpolations are not allowed in declarations.")
elif decl.upper().startswith("DOCTYPE "):
doctype_content = decl[7:].strip()
- doctype = TDocumentType(doctype_content)
+ doctype = TDocumentType(doctype_content, parser_pos=self.get_parser_pos())
self.append_child(doctype)
else:
raise NotImplementedError(
@@ -436,6 +626,8 @@ def reset(self):
self.stack = []
self.placeholders = PlaceholderState()
self.source = None
+ self.sinfo_table = {}
+ self.tcomponent_children = {}
def close(self) -> None:
if self.waiting_for_data():
@@ -449,7 +641,49 @@ def close(self) -> None:
"Parser expects more data, is the template valid html?"
)
if self.stack:
- raise ValueError("Invalid HTML structure: unclosed tags remain.")
+ source = self.get_source()
+ e = ValueError("Invalid HTML structure: unclosed tags remain.")
+ # @TODO: We need to determine which tags this might apply to,
+ # this only applies to components.
+ parent = self.stack[-1]
+ if isinstance(parent, OpenTComponent) and self.has_ambiguous_forward_slash(
+ parent.sinfo
+ ):
+ # CASE: "<{C1} attr={value}/>" -- meant to self-close
+ # Maybe user meant to self-close?
+ starttag = source.format_starttag(parent.start_i_index)
+ e.add_note(
+ f'Did you mean to quote the last attribute or put a space before "/>" for "<{{{starttag}}} .../>"?'
+ )
+ else:
+ # CASE: t"<{C2}><{C1} attr=/>"
+ # Maybe user meant to self-close <{C1} ...>, but closed by leaving <{C2}...> open?
+ # CASE: t"<{C3}><{C2}><{C1} attr=/>"
+ for comp in reversed(
+ self.get_closed_tcomps(parent, recurse_component_children=True)
+ ):
+ if (
+ comp.end_i_index is not None
+ and comp.start_i_index != comp.end_i_index
+ and not source.values_match(
+ comp.start_i_index, comp.end_i_index
+ )
+ ):
+ sinfo = (
+ self.sinfo_table.get(comp.parser_pos)
+ if comp.parser_pos is not None
+ else None
+ )
+ starttag = source.format_starttag(comp.start_i_index)
+ endtag = source.format_endtag(comp.end_i_index)
+ e.add_note(
+ f"Component start tag, <{{{starttag}}}>, and end tag, , have values that do not match."
+ )
+ if self.has_ambiguous_forward_slash(sinfo):
+ e.add_note(
+ f'Did you mean to quote the last attribute or put a space before "/>" for "<{{{starttag}}} .../>"?'
+ )
+ raise e
if not self.placeholders.is_empty:
raise ValueError("Some placeholders were never resolved.")
super().close()
diff --git a/tdom/parser_test.py b/tdom/parser_test.py
index d1650ae0..523b918f 100644
--- a/tdom/parser_test.py
+++ b/tdom/parser_test.py
@@ -4,6 +4,7 @@
from .parser import TemplateParser
from .placeholders import make_placeholder_config
+from .source import FrozenPosition
from .template_utils import TemplateRef
from .tnodes import (
TComment,
@@ -468,6 +469,14 @@ def Component():
_ = TemplateParser.parse(t"<{Component}>")
+def test_unmatched_end_component_tag_error():
+ def Component():
+ pass
+
+ with pytest.raises(ValueError, match="Unexpected closing component tag"):
+ _ = TemplateParser.parse(t"")
+
+
def test_placeholder_collision_avoidance():
config = make_placeholder_config()
# This test is to ensure that our placeholder detection avoids collisions
@@ -602,3 +611,81 @@ def test_extract_with_templated_attr_gt_char(self, Component):
strings=("
Hello, World!
",), i_indexes=()
),
)
+
+
+class TestComponentUnquotedAttrValueWithAmbiguousSlash:
+ @pytest.fixture
+ def comp_maker(self):
+ def maker(suffix=None):
+ def _Comp(children: Template, title: str) -> Template:
+ return children
+
+ if suffix is not None:
+ _Comp.__name__ = f"{_Comp.__name__}__{suffix}"
+ return _Comp
+
+ return maker
+
+ @pytest.fixture
+ def Comp(self):
+ def _Comp(children: Template, title: str) -> Template:
+ return children
+
+ return _Comp
+
+ @pytest.fixture
+ def Comp2(self):
+ def _Comp2(children: Template, title: str) -> Template:
+ return children
+
+ return _Comp2
+
+ def test_comp_unquoted_attr_value_error_root(self, Comp):
+ with pytest.raises(
+ ValueError, match="Did you mean to quote the last attribute"
+ ):
+ _ = TemplateParser.parse(t"<{Comp} title=today/>")
+
+ def test_comp_unquoted_attr_value_error_nested_in_el(self, Comp):
+ with pytest.raises(
+ ValueError, match="Did you mean to quote the last attribute"
+ ):
+ _ = TemplateParser.parse(t"
<{Comp} title=today/>
")
+
+ def test_comp_unquoted_attr_value_error_single_nested_in_comp(self, Comp, Comp2):
+ with pytest.raises(
+ ValueError, match="Did you mean to quote the last attribute"
+ ):
+ _ = TemplateParser.parse(t"<{Comp2}><{Comp} title=today/>")
+
+ def test_comp_unquoted_attr_value_error_double_nested_in_comp(self, comp_maker):
+ Comp1, Comp2, Comp3 = comp_maker("1"), comp_maker("2"), comp_maker("3")
+ with pytest.raises(
+ ValueError, match="Did you mean to quote the last attribute"
+ ):
+ _ = TemplateParser.parse(
+ t"<{Comp2}><{Comp1}><{Comp3} title=today/>"
+ )
+
+
+def PositionComp() -> Template:
+ return t""
+
+
+def test_tnode_parser_position():
+ for tnode_type, fragment in (
+ (TElement, t"
"),
+ (TComment, t""),
+ (TDocumentType, t""),
+ (TComponent, t"<{PositionComp}>"),
+ (TText, t"Just a simple text."),
+ ):
+ tnode = TemplateParser.parse(t"
" + fragment + t"
")
+ assert (
+ isinstance(tnode, TElement)
+ and tnode.tag == "div"
+ and len(tnode.children) == 1
+ )
+ el = tnode.children[0]
+ assert isinstance(el, tnode_type)
+ assert el.parser_pos == FrozenPosition(line=1, offset=len("
"))
diff --git a/tdom/parser_utils.py b/tdom/parser_utils.py
new file mode 100644
index 00000000..f8365871
--- /dev/null
+++ b/tdom/parser_utils.py
@@ -0,0 +1 @@
+type HTMLAttribute = tuple[str, str | None]
diff --git a/tdom/placeholders.py b/tdom/placeholders.py
index 1cf47128..eabe4583 100644
--- a/tdom/placeholders.py
+++ b/tdom/placeholders.py
@@ -61,6 +61,9 @@ class PlaceholderState:
config: PlaceholderConfig = field(default_factory=make_placeholder_config)
"""Collection of currently 'known and active' placeholder indexes."""
+ def copy(self):
+ return PlaceholderState(known=self.known.copy(), config=self.config)
+
@property
def is_empty(self) -> bool:
return len(self.known) == 0
diff --git a/tdom/source.py b/tdom/source.py
new file mode 100644
index 00000000..7ff90726
--- /dev/null
+++ b/tdom/source.py
@@ -0,0 +1,47 @@
+from dataclasses import dataclass
+
+from .parser_utils import HTMLAttribute
+
+
+@dataclass(slots=True, frozen=True)
+class FrozenPosition:
+ "A immutable position in a block of source code."
+
+ line: int = 1
+ " Line of code, starts at 1. "
+ offset: int = 0
+ " Offset from the start of the line, starts at 0. "
+
+
+@dataclass(slots=True)
+class Position:
+ "A position in a block of source code."
+
+ line: int = 1
+ " Line of code, starts at 1. "
+ offset: int = 0
+ " Offset from the start of the line, starts at 0. "
+
+ def freeze(self) -> FrozenPosition:
+ return FrozenPosition(line=self.line, offset=self.offset)
+
+
+@dataclass(frozen=True, slots=True)
+class TagSourceInfo:
+ """
+ Retained tag information from the parsed source.
+
+ @NOTE: These properties DEPEND on the placeholder configuration because
+ they can contain embedded placeholders.
+ """
+
+ starttag_text: str
+ " Entire starttag as parsed, includes placeholders, . "
+ raw_attrs: tuple[HTMLAttribute, ...]
+ " Attrs as parsed, includes placeholders. "
+ startend: bool
+ " Was parsed as startend tag, ie. . "
+ starttag_pos: FrozenPosition
+ " Position of the parser when the element starttag was parsed. "
+ endtag_pos: FrozenPosition | None = None
+ " Position of the parser when the element endtag was parsed. "
diff --git a/tdom/tnodes.py b/tdom/tnodes.py
index 3afb1063..cee6d6da 100644
--- a/tdom/tnodes.py
+++ b/tdom/tnodes.py
@@ -1,6 +1,7 @@
import typing as t
from dataclasses import dataclass, field
+from .source import FrozenPosition
from .template_utils import TemplateRef
@@ -45,6 +46,8 @@ def __str__(self) -> str:
class TText(TNode):
ref: TemplateRef
+ parser_pos: FrozenPosition | None = field(default=None, compare=False)
+
@classmethod
def empty(cls) -> t.Self:
return cls(TemplateRef.empty())
@@ -58,6 +61,8 @@ def literal(cls, text: str) -> t.Self:
class TComment(TNode):
ref: TemplateRef
+ parser_pos: FrozenPosition | None = field(default=None, compare=False)
+
@classmethod
def literal(cls, text: str) -> t.Self:
return cls(TemplateRef.literal(text))
@@ -67,11 +72,15 @@ def literal(cls, text: str) -> t.Self:
class TDocumentType(TNode):
text: str
+ parser_pos: FrozenPosition | None = field(default=None, compare=False)
+
@dataclass(slots=True, frozen=True)
class TFragment(TNode):
children: tuple[TNode, ...] = field(default_factory=tuple)
+ parser_pos: FrozenPosition | None = field(default=None, compare=False)
+
@dataclass(slots=True, frozen=True)
class TElement(TNode):
@@ -79,6 +88,8 @@ class TElement(TNode):
attrs: tuple[TAttribute, ...] = field(default_factory=tuple)
children: tuple[TNode, ...] = field(default_factory=tuple)
+ parser_pos: FrozenPosition | None = field(default=None, compare=False)
+
@dataclass(slots=True, frozen=True)
class TComponent(TNode):
@@ -95,5 +106,7 @@ class TComponent(TNode):
attrs: tuple[TAttribute, ...] = field(default_factory=tuple)
+ parser_pos: FrozenPosition | None = field(default=None, compare=False)
+
type TTag = TElement | TComponent | TFragment