kreuzberg 4.0.0.pre.rc.29 → 4.0.0.rc1

data/vendor/kreuzberg-tesseract/src/mutable_iterator.rs

@@ -1,197 +0,0 @@
-use crate::error::{Result, TesseractError};
-use std::ffi::CStr;
-use std::os::raw::{c_char, c_void};
-use std::sync::Arc;
-use std::sync::Mutex;
-
-use crate::result_iterator::{
-    TessResultIteratorConfidence, TessResultIteratorGetUTF8Text, TessResultIteratorNext,
-    TessResultIteratorSymbolIsDropcap, TessResultIteratorSymbolIsSubscript, TessResultIteratorSymbolIsSuperscript,
-    TessResultIteratorWordFontAttributes, TessResultIteratorWordIsFromDictionary, TessResultIteratorWordIsNumeric,
-    TessResultIteratorWordRecognitionLanguage,
-};
-
-pub struct MutableIterator {
-    handle: Arc<Mutex<*mut c_void>>,
-}
-
-unsafe impl Send for MutableIterator {}
-unsafe impl Sync for MutableIterator {}
-
-impl MutableIterator {
-    /// Creates a new instance of the MutableIterator.
-    ///
-    /// # Arguments
-    ///
-    /// * `handle` - Pointer to the MutableIterator.
-    pub fn new(handle: *mut c_void) -> Self {
-        MutableIterator {
-            handle: Arc::new(Mutex::new(handle)),
-        }
-    }
-
-    /// Gets the UTF-8 text for the current iterator.
-    ///
-    /// # Arguments
-    ///
-    /// * `level` - Level of the text.
-    pub fn get_utf8_text(&self, level: i32) -> Result<String> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        let text_ptr = unsafe { TessResultIteratorGetUTF8Text(*handle, level) };
-        if text_ptr.is_null() {
-            return Err(TesseractError::NullPointerError);
-        }
-        let c_str = unsafe { CStr::from_ptr(text_ptr) };
-        let result = c_str.to_str()?.to_owned();
-        unsafe { TessDeleteText(text_ptr as *mut c_char) };
-        Ok(result)
-    }
-
-    /// Gets the confidence of the current iterator.
-    ///
-    /// # Arguments
-    ///
-    /// * `level` - Level of the confidence.
-    pub fn confidence(&self, level: i32) -> Result<f32> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        Ok(unsafe { TessResultIteratorConfidence(*handle, level) })
-    }
-
-    /// Gets the recognition language of the current iterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns the recognition language as a `String` if successful, otherwise returns an error.
-    pub fn word_recognition_language(&self) -> Result<String> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        let lang_ptr = unsafe { TessResultIteratorWordRecognitionLanguage(*handle) };
-        if lang_ptr.is_null() {
-            return Err(TesseractError::NullPointerError);
-        }
-        let c_str = unsafe { CStr::from_ptr(lang_ptr) };
-        Ok(c_str.to_str()?.to_owned())
-    }
-
-    /// Gets the font attributes of the current iterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns the font attributes as a tuple if successful, otherwise returns an error.
-    pub fn word_font_attributes(&self) -> Result<(bool, bool, bool, bool, bool, bool, i32, i32)> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        let mut is_bold = 0;
-        let mut is_italic = 0;
-        let mut is_underlined = 0;
-        let mut is_monospace = 0;
-        let mut is_serif = 0;
-        let mut is_smallcaps = 0;
-        let mut pointsize = 0;
-        let mut font_id = 0;
-
-        let result = unsafe {
-            TessResultIteratorWordFontAttributes(
-                *handle,
-                &mut is_bold,
-                &mut is_italic,
-                &mut is_underlined,
-                &mut is_monospace,
-                &mut is_serif,
-                &mut is_smallcaps,
-                &mut pointsize,
-                &mut font_id,
-            )
-        };
-
-        if result == 0 {
-            Err(TesseractError::InvalidParameterError)
-        } else {
-            Ok((
-                is_bold != 0,
-                is_italic != 0,
-                is_underlined != 0,
-                is_monospace != 0,
-                is_serif != 0,
-                is_smallcaps != 0,
-                pointsize,
-                font_id,
-            ))
-        }
-    }
-
-    /// Checks if the current word is from the dictionary.
-    ///
-    /// # Returns
-    ///
-    /// Returns `Ok(true)` if the current word is from the dictionary, otherwise returns `Ok(false)`.
-    pub fn word_is_from_dictionary(&self) -> Result<bool> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        Ok(unsafe { TessResultIteratorWordIsFromDictionary(*handle) != 0 })
-    }
-
-    /// Checks if the current word is numeric.
-    ///
-    /// # Returns
-    ///
-    /// Returns `Ok(true)` if the current word is numeric, otherwise returns `Ok(false)`.
-    pub fn word_is_numeric(&self) -> Result<bool> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        Ok(unsafe { TessResultIteratorWordIsNumeric(*handle) != 0 })
-    }
-
-    /// Checks if the current symbol is superscript.
-    ///
-    /// # Returns
-    ///
-    /// Returns `Ok(true)` if the current symbol is superscript, otherwise returns `Ok(false)`.
-    pub fn symbol_is_superscript(&self) -> Result<bool> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        Ok(unsafe { TessResultIteratorSymbolIsSuperscript(*handle) != 0 })
-    }
-
-    /// Checks if the current symbol is subscript.
-    ///
-    /// # Returns
-    ///
-    /// Returns `Ok(true)` if the current symbol is subscript, otherwise returns `Ok(false)`.
-    pub fn symbol_is_subscript(&self) -> Result<bool> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        Ok(unsafe { TessResultIteratorSymbolIsSubscript(*handle) != 0 })
-    }
-
-    /// Checks if the current symbol is dropcap.
-    ///
-    /// # Returns
-    ///
-    /// Returns `Ok(true)` if the current symbol is dropcap, otherwise returns `Ok(false)`.
-    pub fn symbol_is_dropcap(&self) -> Result<bool> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        Ok(unsafe { TessResultIteratorSymbolIsDropcap(*handle) != 0 })
-    }
-
-    /// Gets the next iterator.
-    ///
-    /// # Arguments
-    ///
-    /// * `level` - Level of the iterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns `true` if the next iterator is successful, otherwise returns `false`.
-    pub fn next(&self, level: i32) -> Result<bool> {
-        let handle = self.handle.lock().map_err(|_| TesseractError::MutexError)?;
-        Ok(unsafe { TessResultIteratorNext(*handle, level) != 0 })
-    }
-}
-
-impl Drop for MutableIterator {
-    fn drop(&mut self) {
-        if let Ok(handle) = self.handle.lock() {
-            unsafe { TessResultIteratorDelete(*handle) };
-        }
-    }
-}
-
-unsafe extern "C" {
-    pub fn TessResultIteratorDelete(handle: *mut c_void);
-    pub fn TessDeleteText(text: *mut c_char);
-}

data/vendor/kreuzberg-tesseract/src/page_iterator.rs

@@ -1,253 +0,0 @@
-use crate::TesseractError;
-use crate::enums::{
-    TessOrientation, TessPageIteratorLevel, TessParagraphJustification, TessPolyBlockType, TessTextlineOrder,
-    TessWritingDirection,
-};
-use std::os::raw::{c_float, c_int, c_void};
-use std::sync::Arc;
-use std::sync::Mutex;
-
-pub struct PageIterator {
-    pub handle: Arc<Mutex<*mut c_void>>,
-}
-
-unsafe impl Send for PageIterator {}
-unsafe impl Sync for PageIterator {}
-
-impl PageIterator {
-    /// Creates a new instance of the PageIterator.
-    ///
-    /// # Arguments
-    ///
-    /// * `handle` - Pointer to the PageIterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns the new instance of the PageIterator.
-    pub fn new(handle: *mut c_void) -> Self {
-        PageIterator {
-            handle: Arc::new(Mutex::new(handle)),
-        }
-    }
-
-    /// Begins the iteration.
-    pub fn begin(&self) {
-        let handle = self.handle.lock().unwrap();
-        unsafe { TessPageIteratorBegin(*handle) };
-    }
-
-    /// Gets the next iterator.
-    ///
-    /// # Arguments
-    ///
-    /// * `level` - Level of the iterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns `true` if the next iterator is successful, otherwise returns `false`.
-    pub fn next(&self, level: TessPageIteratorLevel) -> bool {
-        let handle = self.handle.lock().unwrap();
-        unsafe { TessPageIteratorNext(*handle, level as c_int) != 0 }
-    }
-
-    /// Checks if the current iterator is at the beginning of the specified level.
-    ///
-    /// # Arguments
-    ///
-    /// * `level` - Level of the iterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns `true` if the current iterator is at the beginning of the specified level, otherwise returns `false`.
-    pub fn is_at_beginning_of(&self, level: TessPageIteratorLevel) -> bool {
-        let handle = self.handle.lock().unwrap();
-        unsafe { TessPageIteratorIsAtBeginningOf(*handle, level as c_int) != 0 }
-    }
-
-    /// Checks if the current iterator is at the final element of the specified level.
-    ///
-    /// # Arguments
-    ///
-    /// * `level` - Level of the iterator.
-    /// * `element` - Element of the iterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns `true` if the current iterator is at the final element of the specified level, otherwise returns `false`.
-    pub fn is_at_final_element(&self, level: TessPageIteratorLevel, element: TessPageIteratorLevel) -> bool {
-        let handle = self.handle.lock().unwrap();
-        unsafe { TessPageIteratorIsAtFinalElement(*handle, level as c_int, element as c_int) != 0 }
-    }
-
-    /// Gets the bounding box of the current iterator.
-    ///
-    /// # Arguments
-    ///
-    /// * `level` - Level of the bounding box.
-    ///
-    /// # Returns
-    ///
-    /// Returns the bounding box as a tuple if successful, otherwise returns an error.
-    pub fn bounding_box(&self, level: TessPageIteratorLevel) -> Result<(i32, i32, i32, i32), TesseractError> {
-        let mut left = 0;
-        let mut top = 0;
-        let mut right = 0;
-        let mut bottom = 0;
-        let handle = self.handle.lock().unwrap();
-        let result = unsafe {
-            TessPageIteratorBoundingBox(*handle, level as c_int, &mut left, &mut top, &mut right, &mut bottom)
-        };
-        if result == 0 {
-            Err(TesseractError::InvalidParameterError)
-        } else {
-            Ok((left, top, right, bottom))
-        }
-    }
-
-    /// Gets the block type of the current iterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns the block type as a `TessPolyBlockType`.
-    pub fn block_type(&self) -> TessPolyBlockType {
-        let handle = self.handle.lock().unwrap();
-        let block_type = unsafe { TessPageIteratorBlockType(*handle) };
-        unsafe { std::mem::transmute(block_type) }
-    }
-
-    /// Gets the baseline of the current iterator.
-    ///
-    /// # Arguments
-    ///
-    /// * `level` - Level of the baseline.
-    ///
-    /// # Returns
-    ///
-    /// Returns the baseline as a tuple if successful, otherwise returns an error.
-    pub fn baseline(&self, level: i32) -> Result<(i32, i32, i32, i32), TesseractError> {
-        let mut x1 = 0;
-        let mut y1 = 0;
-        let mut x2 = 0;
-        let mut y2 = 0;
-        let handle = self.handle.lock().unwrap();
-        let result = unsafe { TessPageIteratorBaseline(*handle, level, &mut x1, &mut y1, &mut x2, &mut y2) };
-        if result == 0 {
-            Err(TesseractError::InvalidParameterError)
-        } else {
-            Ok((x1, y1, x2, y2))
-        }
-    }
-
-    /// Gets the orientation of the current iterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns the orientation as a tuple if successful, otherwise returns an error.
-    pub fn orientation(
-        &self,
-    ) -> Result<(TessOrientation, TessWritingDirection, TessTextlineOrder, f32), TesseractError> {
-        let mut orientation = 0;
-        let mut writing_direction = 0;
-        let mut textline_order = 0;
-        let mut deskew_angle = 0.0;
-        let handle = self.handle.lock().unwrap();
-        let result = unsafe {
-            TessPageIteratorOrientation(
-                *handle,
-                &mut orientation,
-                &mut writing_direction,
-                &mut textline_order,
-                &mut deskew_angle,
-            )
-        };
-        if result == 0 {
-            Err(TesseractError::InvalidParameterError)
-        } else {
-            Ok((
-                unsafe { std::mem::transmute(orientation) },
-                unsafe { std::mem::transmute(writing_direction) },
-                unsafe { std::mem::transmute(textline_order) },
-                deskew_angle,
-            ))
-        }
-    }
-
-    /// Gets the paragraph information of the current iterator.
-    ///
-    /// # Returns
-    ///
-    /// Returns the paragraph information as a tuple if successful, otherwise returns an error.
-    pub fn paragraph_info(&self) -> Result<(TessParagraphJustification, bool, bool, i32), TesseractError> {
-        let mut justification = 0;
-        let mut is_list_item = false;
-        let mut is_crown = false;
-        let mut first_line_indent = 0;
-        let handle = self.handle.lock().unwrap();
-        let result = unsafe {
-            TessPageIteratorParagraphInfo(
-                *handle,
-                &mut justification,
-                &mut is_list_item,
-                &mut is_crown,
-                &mut first_line_indent,
-            )
-        };
-        if result == 0 {
-            Err(TesseractError::InvalidParameterError)
-        } else {
-            Ok((
-                TessParagraphJustification::from_int(justification),
-                is_list_item,
-                is_crown,
-                first_line_indent,
-            ))
-        }
-    }
-}
-
-impl Drop for PageIterator {
-    fn drop(&mut self) {
-        let handle = self.handle.lock().unwrap();
-        unsafe { TessPageIteratorDelete(*handle) };
-    }
-}
-
-unsafe extern "C" {
-    pub fn TessPageIteratorDelete(handle: *mut c_void);
-    pub fn TessPageIteratorBegin(handle: *mut c_void);
-    pub fn TessPageIteratorNext(handle: *mut c_void, level: c_int) -> c_int;
-    pub fn TessPageIteratorIsAtBeginningOf(handle: *mut c_void, level: c_int) -> c_int;
-    pub fn TessPageIteratorIsAtFinalElement(handle: *mut c_void, level: c_int, element: c_int) -> c_int;
-    pub fn TessPageIteratorBoundingBox(
-        handle: *mut c_void,
-        level: c_int,
-        left: *mut c_int,
-        top: *mut c_int,
-        right: *mut c_int,
-        bottom: *mut c_int,
-    ) -> c_int;
-    pub fn TessPageIteratorBlockType(handle: *mut c_void) -> c_int;
-    pub fn TessPageIteratorBaseline(
-        handle: *mut c_void,
-        level: c_int,
-        x1: *mut c_int,
-        y1: *mut c_int,
-        x2: *mut c_int,
-        y2: *mut c_int,
-    ) -> c_int;
-    pub fn TessPageIteratorOrientation(
-        handle: *mut c_void,
-        orientation: *mut c_int,
-        writing_direction: *mut c_int,
-        textline_order: *mut c_int,
-        deskew_angle: *mut c_float,
-    ) -> c_int;
-    pub fn TessBaseAPIGetIterator(handle: *mut c_void) -> *mut c_void;
-    pub fn TessPageIteratorParagraphInfo(
-        handle: *mut c_void,
-        justification: *mut c_int,
-        is_list_item: *mut bool,
-        is_crown: *mut bool,
-        first_line_indent: *mut c_int,
-    ) -> c_int;
-}