wip

src/main.rs

@@ -44,6 +44,12 @@ mod non_nan_float {
         pub(crate) const fn get(self) -> f32 {
             self.0
         }
+        pub(crate) const fn min(self, other: Self) -> Self {
+            Self(self.0.min(other.0))
+        }
+        pub(crate) const fn max(self, other: Self) -> Self {
+            Self(self.0.max(other.0))
+        }
     }
 
     impl std::hash::Hash for NonNaNF32 {
@@ -1484,10 +1490,95 @@ enum PageItem {
     LineOrRect(LineOrRect),
 }
 
-#[derive(Clone, Debug)]
+#[derive(Copy, Clone, Debug)]
 enum LineOrRect {
-    Line(()),
+    Line(Line),
-    Rect(()),
+    Rect(Rect),
+}
+
+impl LineOrRect {
+    fn width(self) -> f32 {
+        match self {
+            Self::Line(v) => v.width(),
+            Self::Rect(v) => v.width(),
+        }
+    }
+    fn height(self) -> f32 {
+        match self {
+            Self::Line(v) => v.height(),
+            Self::Rect(v) => v.height(),
+        }
+    }
+    fn min_x(self) -> NonNaNF32 {
+        match self {
+            Self::Line(v) => v.min_x(),
+            Self::Rect(v) => v.min_x,
+        }
+    }
+    fn max_x(self) -> NonNaNF32 {
+        match self {
+            Self::Line(v) => v.max_x(),
+            Self::Rect(v) => v.max_x,
+        }
+    }
+    fn min_y(self) -> NonNaNF32 {
+        match self {
+            Self::Line(v) => v.min_y(),
+            Self::Rect(v) => v.min_y,
+        }
+    }
+    fn max_y(self) -> NonNaNF32 {
+        match self {
+            Self::Line(v) => v.max_y(),
+            Self::Rect(v) => v.max_y,
+        }
+    }
+}
+
+#[derive(Copy, Clone, Debug)]
+struct Line {
+    p0_x: NonNaNF32,
+    p0_y: NonNaNF32,
+    p1_x: NonNaNF32,
+    p1_y: NonNaNF32,
+}
+
+impl Line {
+    fn width(self) -> f32 {
+        f32::abs(self.p0_x.get() - self.p1_x.get())
+    }
+    fn height(self) -> f32 {
+        f32::abs(self.p0_y.get() - self.p1_y.get())
+    }
+    fn min_x(self) -> NonNaNF32 {
+        self.p0_x.min(self.p1_x)
+    }
+    fn max_x(self) -> NonNaNF32 {
+        self.p0_x.max(self.p1_x)
+    }
+    fn min_y(self) -> NonNaNF32 {
+        self.p0_y.min(self.p1_y)
+    }
+    fn max_y(self) -> NonNaNF32 {
+        self.p0_y.max(self.p1_y)
+    }
+}
+
+#[derive(Copy, Clone, Debug)]
+struct Rect {
+    min_x: NonNaNF32,
+    max_x: NonNaNF32,
+    min_y: NonNaNF32,
+    max_y: NonNaNF32,
+}
+
+impl Rect {
+    fn width(self) -> f32 {
+        self.max_x.get() - self.min_x.get()
+    }
+    fn height(self) -> f32 {
+        self.max_y.get() - self.min_y.get()
+    }
 }
 
 #[derive(Debug)]
@@ -2039,7 +2130,7 @@ impl Parser {
             Err(
                 e @ ErrorWithNote {
                     error:
-                        ExtractInsnsError::InsnParseError(_) | ExtractInsnsError::PageParseError(_),
+                        ExtractInsnsError::InsnParseError(..) | ExtractInsnsError::PageParseError(..),
                     ..
                 },
             ) => {
@@ -2064,7 +2155,7 @@ impl Parser {
         let page = self.page()?;
         let unprocessed_chars = self.unprocessed_chars()?;
         let ControlFlow::<Infallible>::Continue(()) =
-            page.qt[&self.text_section].range(min_x, max_x, min_y, max_y, |x, y, ch| {
+            page.qt[&self.text_section].range(min_x, max_x, min_y, max_y, |_x, _y, ch| {
                 let PageItem::Char(ch) = ch else {
                     return ControlFlow::Continue(());
                 };
@@ -2119,7 +2210,7 @@ impl Parser {
             max_x,
             start_min_y - fonts.regular()[0].size() * 0.4,
             start_min_y + fonts.regular()[0].size() * 0.6,
-            |x, y, ch| {
+            |_x, _y, ch| {
                 let PageItem::Char(ch) = ch else {
                     return ControlFlow::Continue(());
                 };
@@ -2260,6 +2351,597 @@ impl Parser {
         }
         Ok(Some(retval))
     }
+    fn extract_following_text_lines(
+        &mut self,
+        first_text_line: ParsedTextLine,
+        min_x: f32,
+        max_x: f32,
+        allowed_start_min_y_error: Option<f32>,
+    ) -> Result<Vec<ParsedTextLine>, ExtractInsnsError> {
+        let mut retval = Vec::new();
+        let fonts = first_text_line.fonts;
+        let mut line = Some(first_text_line);
+        while let Some(cur_line) = line {
+            let start_min_y = cur_line.regular_min_y - fonts.regular()[0].line_height();
+            retval.push(cur_line);
+            line = self.extract_text_line(
+                None,
+                start_min_y,
+                min_x,
+                max_x,
+                fonts,
+                0,
+                false,
+                allowed_start_min_y_error,
+            )?;
+        }
+        return Ok(retval);
+    }
+    fn extract_insn_bit_fields(
+        &mut self,
+        mnemonic_lines: &[ParsedTextLine],
+    ) -> Result<Option<InsnBitFields>, ExtractInsnsError> {
+        let mut found_non_affix_line = false;
+        let [.., last_mnemonic_line] = mnemonic_lines else {
+            unreachable!();
+        };
+        let expected_non_affix_line_y = last_mnemonic_line.regular_min_y
+            - if mnemonic_lines.len() > 1 {
+                INSN_BIT_FIELDS_TOP_PAD_HEIGHT2
+            } else {
+                INSN_BIT_FIELDS_TOP_PAD_HEIGHT
+            };
+        let page = self.page().map_err(ExtractInsnsError::Other)?;
+        let _ = page.qt[&self.text_section].range(
+            self.text_section.min_x.get() - 5.0,
+            self.text_section.max_x.get() + 5.0,
+            expected_non_affix_line_y - 5.0,
+            expected_non_affix_line_y + 5.0,
+            |_x, _y, line| {
+                let PageItem::LineOrRect(LineOrRect::Line(line)) = line else {
+                    return ControlFlow::Continue(());
+                };
+                if line.width() > line.height() {
+                    found_non_affix_line = true;
+                    return ControlFlow::Break(());
+                }
+                ControlFlow::Continue(())
+            },
+        );
+        if found_non_affix_line {
+            return self.extract_insn_bit_fields_box(expected_non_affix_line_y);
+        };
+        let prefix_text = self.extract_text_line(
+            None,
+            last_mnemonic_line.regular_min_y - INSN_BIT_FIELDS_PREFIX_TEXT_TOP_PAD_HEIGHT,
+            self.text_section.min_x.get(),
+            self.text_section.max_x.get(),
+            TextLineFonts::InsnBitFieldsAffixTitleFonts,
+            0,
+            true,
+            Some(2.0),
+        )?;
+        let Some(prefix_text) = prefix_text else {
+            return Err(ExtractInsnsError::InsnParseError(
+                "can't find insn prefix bit fields title".into(),
+                Backtrace::capture(),
+            ));
+        };
+        let prefix_text_str = prefix_text.element.inner_text();
+        if prefix_text_str != "Prefix:" {
+            return Err(ExtractInsnsError::InsnParseError(
+                format!("insn prefix bit fields title is not as expected: {prefix_text_str:?}"),
+                Backtrace::capture(),
+            ));
+        }
+        let prefix_bit_fields = self.extract_insn_bit_fields_box(
+            prefix_text.regular_min_y - INSN_BIT_FIELDS_AFFIX_TEXT_TO_BOX_TOP_HEIGHT,
+        )?;
+        let Some(prefix_bit_fields) = prefix_bit_fields else {
+            return Err(ExtractInsnsError::InsnParseError(
+                "can't find insn prefix bit fields".into(),
+                Backtrace::capture(),
+            ));
+        };
+        let suffix_text = self.extract_text_line(
+            None,
+            prefix_bit_fields.box_min_y - INSN_BIT_FIELDS_PREFIX_BOX_BOTTOM_TO_SUFFIX_TEXT_HEIGHT,
+            self.text_section.min_x.get(),
+            self.text_section.max_x.get(),
+            TextLineFonts::InsnBitFieldsAffixTitleFonts,
+            0,
+            true,
+            Some(2.0),
+        )?;
+        let Some(suffix_text) = suffix_text else {
+            return Err(ExtractInsnsError::InsnParseError(
+                "can't find insn suffix bit fields title".into(),
+                Backtrace::capture(),
+            ));
+        };
+        let suffix_text_str = suffix_text.element.inner_text();
+        if suffix_text_str != "Suffix:" {
+            return Err(ExtractInsnsError::InsnParseError(
+                format!("insn suffix bit fields title is not as expected: {suffix_text_str:?}"),
+                Backtrace::capture(),
+            ));
+        }
+        let suffix_bit_fields = self.extract_insn_bit_fields_box(
+            suffix_text.regular_min_y - INSN_BIT_FIELDS_AFFIX_TEXT_TO_BOX_TOP_HEIGHT,
+        )?;
+        let Some(suffix_bit_fields) = suffix_bit_fields else {
+            return Err(ExtractInsnsError::InsnParseError(
+                "can't find insn suffix bit fields".into(),
+                Backtrace::capture(),
+            ));
+        };
+        return Ok(Some(InsnBitFields {
+            prefix: Some(InsnBitFieldsPrefix {
+                box_min_x: prefix_bit_fields.box_min_x,
+                box_min_y: prefix_bit_fields.box_min_y,
+                box_max_x: prefix_bit_fields.box_max_x,
+                box_max_y: prefix_bit_fields.box_max_y,
+                prefix_text: prefix_text,
+                fields: prefix_bit_fields.fields,
+                suffix_text: suffix_text,
+            }),
+            box_min_x: suffix_bit_fields.box_min_x,
+            box_min_y: suffix_bit_fields.box_min_y,
+            box_max_x: suffix_bit_fields.box_max_x,
+            box_max_y: suffix_bit_fields.box_max_y,
+            fields: suffix_bit_fields.fields,
+        }));
+    }
+    fn extract_insn_bit_fields_box(
+        &mut self,
+        expected_box_max_y: f32,
+    ) -> Result<Option<InsnBitFields>, ExtractInsnsError> {
+        let mut h_lines = Vec::new();
+        let mut v_lines = Vec::new();
+        let page = self.page().map_err(ExtractInsnsError::Other)?;
+        let ControlFlow::<Infallible>::Continue(()) = page.qt[&self.text_section].range(
+            self.text_section.min_x.get() - 5.0,
+            self.text_section.max_x.get() + 5.0,
+            expected_box_max_y - INSN_BIT_FIELDS_BOX_HEIGHT - 5.0,
+            expected_box_max_y + 5.0,
+            |_x, _y, line| {
+                let PageItem::LineOrRect(LineOrRect::Line(line)) = *line else {
+                    return ControlFlow::Continue(());
+                };
+                if line.width() > line.height() {
+                    h_lines.push(line);
+                } else {
+                    v_lines.push(line);
+                }
+                ControlFlow::Continue(())
+            },
+        );
+        h_lines.sort_by_key(|line| line.min_y());
+        v_lines.sort_by_key(|line| line.min_x());
+        for i in (0..v_lines.len().saturating_sub(1)).rev() {
+            if f32::abs(v_lines[i].min_x().get() - v_lines[i + 1].min_x().get()) < 0.5 {
+                v_lines.remove(i + 1); // remove duplicates
+            }
+        }
+        if h_lines.is_empty() && v_lines.is_empty() {
+            return Ok(None);
+        }
+        let [bottom_line, top_line] = &*h_lines else {
+            return Err(ExtractInsnsError::InsnParseError(
+                format!(
+                    "instruction bit fields box has wrong number of horizontal lines:\n{h_lines:?}"
+                ),
+                Backtrace::capture(),
+            ));
+        };
+        let [leftmost_line, .., rightmost_line] = &*v_lines else {
+            return Err(ExtractInsnsError::InsnParseError(
+                format!("instruction bit fields box has too few vertical lines:\n{v_lines:?}"),
+                Backtrace::capture(),
+            ));
+        };
+        let box_min_x = leftmost_line.min_x().get();
+        let box_max_x = rightmost_line.min_x().get();
+        let box_min_y = bottom_line.min_y().get();
+        let box_max_y = top_line.max_y().get();
+        let box_mid_y = (box_min_y + box_max_y) * 0.5;
+        println!("bottom_line={bottom_line:?}");
+        println!("top_line={top_line:?}");
+        println!("{v_lines:?}");
+        let mut fields = Vec::new();
+        for i in 0..v_lines.len() - 1 {
+            let left_line = v_lines[i];
+            let right_line = v_lines[i + 1];
+            let field_box_min_x = left_line.max_x().get();
+            let field_box_max_x = right_line.min_x().get();
+            let bit_field_name_start_min_y = box_mid_y + 3.288;
+            let bit_field_name = self.extract_text_line(
+                None,
+                bit_field_name_start_min_y,
+                field_box_min_x,
+                field_box_max_x,
+                TextLineFonts::InsnBitFieldNameFonts,
+                0,
+                true,
+                Some(0.4),
+            )?;
+            let Some(bit_field_name) = bit_field_name else {
+                return Err(ExtractInsnsError::InsnParseError(
+                    format!(
+                        "instruction bit field name not found:\n\
+                start_min_y={bit_field_name_start_min_y} \
+                    field_box_min_x={field_box_min_x} \
+                    field_box_max_x={field_box_max_x}"
+                    ),
+                    Backtrace::capture(),
+                ));
+            };
+            let bit_field_number_start_min_y = box_min_y + 3.487;
+            let bit_number = self.extract_text_line(
+                None,
+                bit_field_number_start_min_y,
+                field_box_min_x,
+                field_box_max_x,
+                TextLineFonts::InsnBitFieldBitNumberFonts,
+                0,
+                true,
+                None,
+            )?;
+            let Some(bit_number) = bit_number else {
+                return Err(ExtractInsnsError::InsnParseError(
+                    format!(
+                        "instruction bit field bit number not found:\n\
+                    start_min_y={bit_field_number_start_min_y} \
+                    field_box_min_x={field_box_min_x} \
+                    field_box_max_x={field_box_max_x}"
+                    ),
+                    Backtrace::capture(),
+                ));
+            };
+            fields.push(InsnBitField {
+                box_min_x: field_box_min_x,
+                box_max_x: field_box_max_x,
+                name: bit_field_name,
+                bit_number: bit_number,
+            });
+        }
+        return Ok(Some(InsnBitFields {
+            prefix: None,
+            box_min_x,
+            box_min_y,
+            box_max_x,
+            box_max_y,
+            fields,
+        }));
+    }
+    fn extract_insn_header_mnemonics_and_bit_fields(
+        &mut self,
+        start_min_y: f32,
+        header_start_char: Option<Char>,
+    ) -> Result<Option<InsnHeader>, ExtractInsnsError> {
+        assert!(
+            header_start_char
+                .as_ref()
+                .is_none_or(|v| v.font == Font::InsnHeader)
+        );
+        let Some(header_line) = self.extract_text_line(
+            header_start_char,
+            start_min_y,
+            self.text_section.min_x.get(),
+            self.text_section.max_x.get(),
+            TextLineFonts::InsnHeaderFonts,
+            0,
+            true,
+            Some(6.0),
+        )?
+        else {
+            return Ok(None);
+        };
+        println!("found header line:\n{header_line}");
+        let header_lines = self.extract_following_text_lines(
+            header_line,
+            self.text_section.min_x.get(),
+            self.text_section.max_x.get(),
+            Some(1.5),
+        )?;
+        println!("insn header lines:");
+        for header_line in &header_lines {
+            println!("{header_line}");
+        }
+        let [.., last_header_line] = &*header_lines else {
+            unreachable!();
+        };
+        let Some(mnemonic_start_char) = self
+            .find_top_left_char_in_range(
+                self.text_section.min_x.get() - 5.0,
+                self.text_section.max_x.get() + 5.0,
+                last_header_line.regular_min_y - 50.0,
+                last_header_line.regular_min_y - 5.0,
+                false,
+            )
+            .map_err(ExtractInsnsError::Other)?
+        else {
+            return Err(ExtractInsnsError::InsnParseError(
+                "can't find insn mnemonic text line".into(),
+                Backtrace::capture(),
+            ));
+        };
+        let mnemonic_start_char_min_y = mnemonic_start_char.min_y.get();
+        let Some(mnemonic_line) = self.extract_text_line(
+            Some(mnemonic_start_char),
+            mnemonic_start_char_min_y,
+            self.text_section.min_x.get(),
+            self.text_section.max_x.get(),
+            TextLineFonts::InsnMnemonicFonts,
+            0,
+            true,
+            None,
+        )?
+        else {
+            return Err(ExtractInsnsError::InsnParseError(
+                "can't find insn mnemonic text line".into(),
+                Backtrace::capture(),
+            ));
+        };
+        let mnemonic_line_first_char_min_x = mnemonic_line.chars[0].min_x.get();
+        let mnemonic_lines = self.extract_following_text_lines(
+            mnemonic_line,
+            mnemonic_line_first_char_min_x,
+            self.text_section.max_x.get(),
+            None,
+        )?;
+        println!("insn mnemonic lines:");
+        for mnemonic_line in &mnemonic_lines {
+            println!("{mnemonic_line}");
+        }
+        let Some(insn_bit_fields) = self.extract_insn_bit_fields(&mnemonic_lines)? else {
+            return Err(ExtractInsnsError::InsnParseError(
+                "can't find insn bit fields".into(),
+                Backtrace::capture(),
+            ));
+        };
+        println!("{insn_bit_fields}");
+        return Ok(Some(InsnHeader {
+            header_lines,
+            mnemonic_lines,
+            bit_fields: insn_bit_fields,
+        }));
+    }
+    fn extract_insn_sp_regs_altered(
+        &mut self,
+        mut sp_regs_altered_text: ParsedTextLine,
+    ) -> Result<InsnSpRegsAltered, ExtractInsnsError> {
+        sp_regs_altered_text.preceding_blank_lines = 0;
+        let fonts = TextLineFonts::InsnDescFonts;
+        let column_min_x = sp_regs_altered_text.chars[0].min_x.get();
+        let Some(table_header_reg_char) = self
+            .find_top_left_char_in_range(
+                column_min_x - 1.0,
+                column_min_x + INSN_SP_REGS_ALTERED_FIELDS_COLUMN_X - 1.0,
+                sp_regs_altered_text.regular_min_y - 30.0,
+                sp_regs_altered_text.regular_min_y - 5.0,
+                false,
+            )
+            .map_err(ExtractInsnsError::Other)?
+        else {
+            return Err(ExtractInsnsError::InsnParseError(
+                "can't find special registers altered table's register-column's header".into(),
+                Backtrace::capture(),
+            ));
+        };
+        const KNOWN_SPECIAL_TEXTS: &[&str] = &[
+            "None",
+            "Dependent on the system service",
+            "See above.",
+            "See Table 5.1",
+        ];
+        match &*table_header_reg_char.text {
+            "R" => {}
+            text if KNOWN_SPECIAL_TEXTS.iter().any(|i| text == &i[..1]) => {
+                let start_min_y = table_header_reg_char.min_y.get();
+                let special_text = self.extract_text_line(
+                    Some(table_header_reg_char),
+                    start_min_y,
+                    column_min_x,
+                    self.text_section.max_x.get(),
+                    fonts,
+                    0,
+                    true,
+                    None,
+                )?;
+                let special_text = match special_text {
+                    Some(special_text)
+                        if KNOWN_SPECIAL_TEXTS.contains(&&*special_text.element.text) =>
+                    {
+                        special_text
+                    }
+                    _ => return Err(ExtractInsnsError::Other(
+                        format!(
+                            "can't find special-registers-altered special-text:\n{special_text:?}"
+                        )
+                        .into(),
+                    )),
+                };
+                let final_regular_min_y = special_text.regular_min_y;
+                return Ok(InsnSpRegsAltered {
+                    sp_regs_altered_text,
+                    special_text: Some(special_text),
+                    table_header_reg: None,
+                    table_header_fields: None,
+                    entries: vec![],
+                    final_regular_min_y,
+                });
+            }
+            text => {
+                return Err(ExtractInsnsError::InsnParseError(
+                    format!(
+                        "unknown special-registers-altered special-text start character: {text:?}"
+                    ),
+                    Backtrace::capture(),
+                ));
+            }
+        }
+        let Some(table_header_fields_char) = self
+            .find_top_left_char_in_range(
+                column_min_x + INSN_SP_REGS_ALTERED_FIELDS_COLUMN_X - 10.0,
+                column_min_x + INSN_SP_REGS_ALTERED_FIELDS_CONDS_SPLIT_X,
+                table_header_reg_char.min_y.get() - 5.0,
+                table_header_reg_char.min_y.get() + 5.0,
+                false,
+            )
+            .map_err(ExtractInsnsError::Other)?
+        else {
+            return Err(ExtractInsnsError::Other(
+                "can't find special registers altered table's fields-column's header".into(),
+            ));
+        };
+        if table_header_fields_char.text != "F" {
+            return Err(ExtractInsnsError::Other(
+                format!(
+                    "can't find special registers altered table's fields-column's header:\n\
+            table_header_fields_char={table_header_fields_char:?}"
+                )
+                .into(),
+            ));
+        }
+        let columns_x_bounds = [
+            (
+                table_header_reg_char.min_x.get(),
+                table_header_fields_char.min_x.get() - 1.0,
+            ),
+            (
+                table_header_fields_char.min_x.get(),
+                column_min_x + INSN_SP_REGS_ALTERED_FIELDS_CONDS_SPLIT_X,
+            ),
+            (
+                column_min_x + INSN_SP_REGS_ALTERED_FIELDS_CONDS_SPLIT_X,
+                self.text_section.max_x.get(),
+            ),
+        ];
+        let start_min_y = table_header_reg_char.min_y.get();
+        let Some(table_header_reg) = self.extract_text_line(
+            Some(table_header_reg_char),
+            start_min_y,
+            columns_x_bounds[0].0,
+            columns_x_bounds[0].1,
+            fonts,
+            0,
+            false,
+            None,
+        )?
+        else {
+            return Err(ExtractInsnsError::Other(
+                "can't find special registers altered table's register-column's header".into(),
+            ));
+        };
+        let table_header_reg_text = table_header_reg.element.inner_text();
+        if table_header_reg_text != "Register" {
+            return Err(ExtractInsnsError::Other(
+                format!(
+                    "can't find special registers altered table's register-column's header:\n\
+            table_header_reg_text={table_header_reg_text:?}"
+                )
+                .into(),
+            ));
+        }
+        let start_min_y = table_header_fields_char.min_y.get();
+        let Some(table_header_fields) = self.extract_text_line(
+            Some(table_header_fields_char),
+            start_min_y,
+            columns_x_bounds[1].0,
+            columns_x_bounds[1].1,
+            fonts,
+            0,
+            false,
+            None,
+        )?
+        else {
+            return Err(ExtractInsnsError::Other(
+                "can't find special registers altered table's fields-column's header".into(),
+            ));
+        };
+        let table_header_fields_text = table_header_fields.element.inner_text();
+        if table_header_reg_text != "Field(s)" {
+            return Err(ExtractInsnsError::Other(
+                format!(
+                    "can't find special registers altered table's fields-column's header:\n\
+                    table_header_fields_text={table_header_fields_text:?}"
+                )
+                .into(),
+            ));
+        }
+        let mut regular_min_y = table_header_reg.regular_min_y;
+        let mut entries = Vec::new();
+        let mut cur_reg = None;
+        let mut cur_fields = Vec::new();
+        let mut cur_conds = Vec::new();
+        loop {
+            let mut row = [None, None, None];
+            let mut next_regular_min_y = None;
+            for (i, (min_x, max_x)) in columns_x_bounds.into_iter().enumerate() {
+                row[i] = self.extract_text_line(
+                    None,
+                    regular_min_y - fonts.regular()[0].line_height(),
+                    min_x,
+                    max_x,
+                    fonts,
+                    0,
+                    true,
+                    Some(2.0),
+                )?;
+                if let Some(cell) = &row[i]
+                    && next_regular_min_y.is_none()
+                {
+                    next_regular_min_y = Some(cell.regular_min_y);
+                }
+            }
+            match next_regular_min_y {
+                Some(v) => regular_min_y = v,
+                None => break,
+            }
+            let [cur_reg_cell, cur_fields_cell, cur_conds_cell] = row;
+            if cur_reg_cell.is_none() {
+                if cur_reg.is_none() {
+                    return Err(ExtractInsnsError::Other(
+                        "can't find special registers altered table's first register".into(),
+                    ));
+                }
+                cur_fields.extend(cur_fields_cell);
+                cur_conds.extend(cur_conds_cell);
+                continue;
+            }
+            if let Some(cur_reg) = cur_reg {
+                entries.push(InsnSpRegsAlteredEntry {
+                    reg: cur_reg,
+                    fields: cur_fields,
+                    conds: cur_conds,
+                });
+                cur_fields = Vec::new();
+                cur_conds = Vec::new();
+            }
+            cur_reg = cur_reg_cell;
+            cur_fields.extend(cur_fields_cell);
+            cur_conds.extend(cur_conds_cell);
+        }
+        let Some(cur_reg) = cur_reg else {
+            return Err(ExtractInsnsError::Other(
+                "can't find special registers altered table's first register".into(),
+            ));
+        };
+        entries.push(InsnSpRegsAlteredEntry {
+            reg: cur_reg,
+            fields: cur_fields,
+            conds: cur_conds,
+        });
+        return Ok(InsnSpRegsAltered {
+            sp_regs_altered_text: sp_regs_altered_text,
+            special_text: None,
+            table_header_reg: Some(table_header_reg),
+            table_header_fields: Some(table_header_fields),
+            entries,
+            final_regular_min_y: regular_min_y,
+        });
+    }
     /*fn extract_insn(&mut self, header_start_char: Char) -> Result<Insn, ExtractInsnsError> {
         assert_eq!(header_start_char.font, Font::InsnHeader);
         println!("{header_start_char:?}");
@@ -2401,57 +3083,192 @@ impl Parser {
     }
 }
 
-#[derive(Clone, Debug, Default)]
+#[derive(Clone, Debug)]
 struct MyDevice {
+    page_num: u32,
     qt: Rc<RefCell<BTreeMap<TextSection, QuadTree<PageItem>>>>,
     unprocessed_non_text: Rc<RefCell<IndexSet<LineOrRect>>>,
 }
 
+impl MyDevice {
+    fn new(page_num: u32) -> Self {
+        Self {
+            page_num,
+            qt: Default::default(),
+            unprocessed_non_text: Default::default(),
+        }
+    }
+    fn path(&mut self, path: &mupdf::Path, cmt: mupdf::Matrix) {
+        enum Walker {
+            Empty,
+            Moved { x: f32, y: f32 },
+            Line(Line),
+            Rect { x1: f32, y1: f32, x2: f32, y2: f32 },
+            NotRecognized,
+        }
+        fn new_line(p0_x: f32, p0_y: f32, p1_x: f32, p1_y: f32) -> Option<Line> {
+            Some(Line {
+                p0_x: NonNaNF32::new(p0_x)?,
+                p0_y: NonNaNF32::new(p0_y)?,
+                p1_x: NonNaNF32::new(p1_x)?,
+                p1_y: NonNaNF32::new(p1_y)?,
+            })
+        }
+        impl mupdf::PathWalker for Walker {
+            fn move_to(&mut self, x: f32, y: f32) {
+                *self = match *self {
+                    Walker::Empty | Walker::Moved { .. } => Walker::Moved { x, y },
+                    Walker::Line(_) | Walker::Rect { .. } | Walker::NotRecognized => {
+                        Walker::NotRecognized
+                    }
+                };
+            }
+            fn line_to(&mut self, x: f32, y: f32) {
+                *self = match *self {
+                    Walker::Empty => Walker::NotRecognized,
+                    Walker::Moved { x: p0_x, y: p0_y } => new_line(p0_x, p0_y, x, y)
+                        .map(Walker::Line)
+                        .unwrap_or(Walker::NotRecognized),
+                    Walker::Line(_) | Walker::Rect { .. } | Walker::NotRecognized => {
+                        Walker::NotRecognized
+                    }
+                };
+            }
+            fn curve_to(&mut self, _cx1: f32, _cy1: f32, _cx2: f32, _cy2: f32, _ex: f32, _ey: f32) {
+                *self = Walker::NotRecognized;
+            }
+            fn close(&mut self) {}
+            fn rect(&mut self, x1: f32, y1: f32, x2: f32, y2: f32) {
+                *self = match *self {
+                    Walker::Empty => Walker::Rect { x1, y1, x2, y2 },
+                    Walker::Moved { .. }
+                    | Walker::Line(..)
+                    | Walker::Rect { .. }
+                    | Walker::NotRecognized => Walker::NotRecognized,
+                };
+            }
+        }
+        let mut walker = Walker::Empty;
+        let Ok(()) = path.walk(&mut walker) else {
+            return;
+        };
+        let component = match walker {
+            Walker::Empty | Walker::Moved { .. } | Walker::NotRecognized => return,
+            Walker::Line(Line {
+                p0_x,
+                p0_y,
+                p1_x,
+                p1_y,
+            }) => {
+                let mupdf::Point { x: p0_x, y: p0_y } = mupdf::Point {
+                    x: p0_x.get(),
+                    y: p0_y.get(),
+                }
+                .transform(&cmt);
+                let mupdf::Point { x: p1_x, y: p1_y } = mupdf::Point {
+                    x: p1_x.get(),
+                    y: p1_y.get(),
+                }
+                .transform(&cmt);
+                let Some(line) = new_line(p0_x, p0_y, p1_x, p1_y) else {
+                    return;
+                };
+                LineOrRect::Line(line)
+            }
+            Walker::Rect { x1, y1, x2, y2 } => {
+                let p1 = mupdf::Point { x: x1, y: y1 }.transform(&cmt);
+                let p2 = mupdf::Point { x: x2, y: y1 }.transform(&cmt);
+                let p3 = mupdf::Point { x: x2, y: y2 }.transform(&cmt);
+                let p4 = mupdf::Point { x: x1, y: y2 }.transform(&cmt);
+                let min_x = NonNaNF32::new(p1.x.min(p2.x).min(p3.x).min(p4.x));
+                let max_x = NonNaNF32::new(p1.x.max(p2.x).max(p3.x).max(p4.x));
+                let min_y = NonNaNF32::new(p1.y.min(p2.y).min(p3.y).min(p4.y));
+                let max_y = NonNaNF32::new(p1.y.max(p2.y).max(p3.y).max(p4.y));
+                let (Some(min_x), Some(max_x), Some(min_y), Some(max_y)) =
+                    (min_x, max_x, min_y, max_y)
+                else {
+                    return;
+                };
+                LineOrRect::Rect(Rect {
+                    min_x,
+                    max_x,
+                    min_y,
+                    max_y,
+                })
+            }
+        };
+        if component.width() > 100.0
+            && component.min_x().get() < COLUMN_SPLIT_X - 10.0
+            && component.max_x().get() > COLUMN_SPLIT_X + 10.0
+        {
+            println!("wide component: {component:?}");
+        } else {
+            println!("component: {component:?}");
+        }
+        let text_section = TextSection::for_position(
+            self.page_num,
+            (component.min_x().get() + component.max_x().get()) * 0.5,
+            (component.min_y().get() + component.max_y().get()) * 0.5,
+        );
+        if let Some(text_section) = text_section {
+            self.qt
+                .borrow_mut()
+                .entry(text_section)
+                .or_default()
+                .insert(
+                    component.min_x().get(),
+                    component.min_y().get(),
+                    PageItem::LineOrRect(component),
+                );
+        }
+    }
+}
+
 impl mupdf::NativeDevice for MyDevice {
     fn fill_path(
         &mut self,
         path: &mupdf::Path,
-        even_odd: bool,
+        _even_odd: bool,
         cmt: mupdf::Matrix,
-        color_space: &mupdf::Colorspace,
+        _color_space: &mupdf::Colorspace,
-        color: &[f32],
+        _color: &[f32],
-        alpha: f32,
+        _alpha: f32,
-        cp: mupdf::ColorParams,
+        _cp: mupdf::ColorParams,
     ) {
-        // TODO
+        self.path(path, cmt);
     }
 
     fn stroke_path(
         &mut self,
         path: &mupdf::Path,
-        stroke_state: &mupdf::StrokeState,
+        _stroke_state: &mupdf::StrokeState,
         cmt: mupdf::Matrix,
-        color_space: &mupdf::Colorspace,
+        _color_space: &mupdf::Colorspace,
-        color: &[f32],
+        _color: &[f32],
-        alpha: f32,
+        _alpha: f32,
-        cp: mupdf::ColorParams,
+        _cp: mupdf::ColorParams,
     ) {
-        // TODO
+        self.path(path, cmt);
     }
 
     fn clip_path(
         &mut self,
         path: &mupdf::Path,
-        even_odd: bool,
+        _even_odd: bool,
         cmt: mupdf::Matrix,
-        scissor: mupdf::Rect,
+        _scissor: mupdf::Rect,
     ) {
-        // TODO
+        self.path(path, cmt);
     }
 
     fn clip_stroke_path(
         &mut self,
         path: &mupdf::Path,
-        stroke_state: &mupdf::StrokeState,
+        _stroke_state: &mupdf::StrokeState,
         cmt: mupdf::Matrix,
-        scissor: mupdf::Rect,
+        _scissor: mupdf::Rect,
     ) {
-        // TODO
+        self.path(path, cmt);
     }
 }
 
@@ -2528,12 +3345,13 @@ impl Page {
         page_num: u32,
         page: mupdf::Page,
     ) -> Result<Self, Box<dyn std::error::Error>> {
-        let device = MyDevice::default();
+        let device = MyDevice::new(page_num);
         page.run(
             &mupdf::Device::from_native(device.clone())?,
             &mupdf::Matrix::IDENTITY,
         )?;
         let MyDevice {
+            page_num,
             qt,
             unprocessed_non_text,
         } = device;