natural-pdf 0.2.12__py3-none-any.whl → 0.2.15__py3-none-any.whl

natural_pdf/vision/template_matching.py

@@ -0,0 +1,209 @@
+"""Pure NumPy template matching implementation"""
+
+from dataclasses import dataclass
+from typing import List, Optional, Tuple
+
+import numpy as np
+
+
+@dataclass
+class TemplateMatch:
+    """Result of template matching"""
+
+    bbox: Tuple[int, int, int, int]  # x0, y0, x1, y1
+    score: float  # 0-1, higher is better
+
+
+class TemplateMatcher:
+    """Pure NumPy template matching implementation"""
+
+    def __init__(self, method: str = "zncc"):
+        """
+        Args:
+            method: Matching method
+                - "zncc": Zero-mean Normalized Cross-Correlation (default, recommended)
+                - "ncc": Normalized Cross-Correlation
+                - "ssd": Sum of Squared Differences
+        """
+        self.method = method
+
+    def match_template(
+        self,
+        image: np.ndarray,
+        template: np.ndarray,
+        step: int = 1,
+        mask_threshold: Optional[float] = None,
+    ) -> np.ndarray:
+        """
+        Compute similarity map between image and template.
+
+        Args:
+            image: Target image (grayscale, normalized 0-1)
+            template: Template to search for (grayscale, normalized 0-1)
+            step: Step size for sliding window (1 = pixel perfect, >1 = faster)
+            mask_threshold: If provided, pixels >= this value in template are masked (ignored).
+                           Useful for ignoring white backgrounds (e.g., 0.95 for near-white)
+
+        Returns:
+            2D array of match scores
+        """
+        if self.method == "zncc":
+            return self._zncc(image, template, step, mask_threshold)
+        elif self.method == "ncc":
+            return self._ncc(image, template, step, mask_threshold)
+        elif self.method == "ssd":
+            return self._ssd(image, template, step, mask_threshold)
+        else:
+            # Default to zncc
+            return self._zncc(image, template, step, mask_threshold)
+
+    def _zncc(
+        self,
+        image: np.ndarray,
+        template: np.ndarray,
+        step: int = 1,
+        mask_threshold: Optional[float] = None,
+    ) -> np.ndarray:
+        """Zero-mean Normalized Cross-Correlation - most robust"""
+        h, w = template.shape
+        img_h, img_w = image.shape
+
+        out_h = (img_h - h) // step + 1
+        out_w = (img_w - w) // step + 1
+        result = np.zeros((out_h, out_w))
+
+        # Create mask if threshold provided
+        if mask_threshold is not None:
+            mask = template < mask_threshold  # True for pixels to keep
+            if np.sum(mask) == 0:
+                # All pixels are masked - return zeros
+                return result
+        else:
+            mask = np.ones_like(template, dtype=bool)
+
+        # Precompute template statistics on non-masked pixels
+        masked_template = template[mask]
+        if len(masked_template) == 0:
+            return result
+
+        template_mean = np.mean(masked_template)
+        template_centered = np.zeros_like(template)
+        template_centered[mask] = template[mask] - template_mean
+        template_std = np.sqrt(np.sum(template_centered[mask] ** 2))
+
+        # Handle uniform template case
+        if template_std == 0:
+            # Template has no variation - fall back to checking if means match
+            for i in range(out_h):
+                for j in range(out_w):
+                    y = i * step
+                    x = j * step
+                    window = image[y : y + h, x : x + w]
+                    window_masked = window[mask]
+                    window_mean = np.mean(window_masked)
+                    window_std = np.std(window_masked)
+
+                    # Perfect match if window also has same mean and no variation
+                    if abs(window_mean - template_mean) < 0.01 and window_std < 0.01:
+                        result[i, j] = 1.0
+            return result
+
+        for i in range(out_h):
+            for j in range(out_w):
+                y = i * step
+                x = j * step
+                window = image[y : y + h, x : x + w]
+
+                # Apply mask to window
+                window_masked = window[mask]
+                window_mean = np.mean(window_masked)
+                window_centered = np.zeros_like(window)
+                window_centered[mask] = window[mask] - window_mean
+                window_std = np.sqrt(np.sum(window_centered[mask] ** 2))
+
+                if window_std > 0:
+                    correlation = np.sum(window_centered[mask] * template_centered[mask])
+                    result[i, j] = correlation / (template_std * window_std)
+
+        return np.clip(result, -1, 1)
+
+    def _ncc(
+        self,
+        image: np.ndarray,
+        template: np.ndarray,
+        step: int = 1,
+        mask_threshold: Optional[float] = None,
+    ) -> np.ndarray:
+        """Normalized Cross-Correlation"""
+        h, w = template.shape
+        img_h, img_w = image.shape
+
+        out_h = (img_h - h) // step + 1
+        out_w = (img_w - w) // step + 1
+        result = np.zeros((out_h, out_w))
+
+        # Create mask if threshold provided
+        if mask_threshold is not None:
+            mask = template < mask_threshold  # True for pixels to keep
+            if np.sum(mask) == 0:
+                return result
+        else:
+            mask = np.ones_like(template, dtype=bool)
+
+        template_norm = np.sqrt(np.sum(template[mask] ** 2))
+        if template_norm == 0:
+            return result
+
+        for i in range(out_h):
+            for j in range(out_w):
+                y = i * step
+                x = j * step
+                window = image[y : y + h, x : x + w]
+
+                window_norm = np.sqrt(np.sum(window[mask] ** 2))
+                if window_norm > 0:
+                    correlation = np.sum(window[mask] * template[mask])
+                    result[i, j] = correlation / (template_norm * window_norm)
+
+        return result
+
+    def _ssd(
+        self,
+        image: np.ndarray,
+        template: np.ndarray,
+        step: int = 1,
+        mask_threshold: Optional[float] = None,
+    ) -> np.ndarray:
+        """Sum of Squared Differences - converted to similarity score"""
+        h, w = template.shape
+        img_h, img_w = image.shape
+
+        out_h = (img_h - h) // step + 1
+        out_w = (img_w - w) // step + 1
+        result = np.zeros((out_h, out_w))
+
+        # Create mask if threshold provided
+        if mask_threshold is not None:
+            mask = template < mask_threshold  # True for pixels to keep
+            if np.sum(mask) == 0:
+                return result
+        else:
+            mask = np.ones_like(template, dtype=bool)
+
+        # Number of valid pixels for normalization
+        n_valid = np.sum(mask)
+        if n_valid == 0:
+            return result
+
+        for i in range(out_h):
+            for j in range(out_w):
+                y = i * step
+                x = j * step
+                window = image[y : y + h, x : x + w]
+
+                # Only compute SSD on non-masked pixels
+                diff = window - template
+                ssd = np.sum((diff[mask]) ** 2) / n_valid
+                result[i, j] = 1.0 / (1.0 + ssd)  # Convert to similarity
+
+        return result

{natural_pdf-0.2.12.dist-info → natural_pdf-0.2.15.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: natural-pdf
-Version: 0.2.12
+Version: 0.2.15
 Summary: A more intuitive interface for working with PDFs
 Author-email: Jonathan Soma <jonathan.soma@gmail.com>
 License-Expression: MIT

{natural_pdf-0.2.12.dist-info → natural_pdf-0.2.15.dist-info}/RECORD

@@ -26,7 +26,7 @@ natural_pdf/classification/results.py,sha256=5ha77CxK0GYwkBMJbvUBZkBjsL5GpOveIZD
 natural_pdf/collections/mixins.py,sha256=Se2C5AcpP9B5E0d0pIrey6-f_P32tAXTK4M7666MNj0,5688
 natural_pdf/core/__init__.py,sha256=QC8H4M3KbXwMFiQORZ0pdPlzx1Ix6oKKQSS7Ib2KEaA,38
 natural_pdf/core/element_manager.py,sha256=KPuKM7SstfErTkRnGq4vrgE0Tv8iazN13Jp7yAXGKso,55575
-natural_pdf/core/highlighting_service.py,sha256=7on8nErhi50CEH2L4XzGIZ6tIqZtMzmmFlp-2lmwnYE,68856
+natural_pdf/core/highlighting_service.py,sha256=wEV-koqHoHf7S3wZ3j8D2L-ucGp3Nd0YhhStz9yqeLc,70406
 natural_pdf/core/page.py,sha256=Pid5hqVjcyX-gcCzxCJ62k6AQhNbUMNM_5QmEcylIjM,155264
 natural_pdf/core/page_collection.py,sha256=IjdFq9q0D0P6ZKWInf0H25rLzxfMb7RsUXucogkhNkU,63169
 natural_pdf/core/page_groupby.py,sha256=V2e_RNlHaasUzYm2h2vNJI7_aV_fl3_pg7kU3F2j0z8,8218
@@ -39,13 +39,13 @@ natural_pdf/describe/elements.py,sha256=3Y541z5TQ2obrfZFiFi1YQMsCt3oYrhMHpD5j1tu
 natural_pdf/describe/mixin.py,sha256=rkX14aGrSz7Jvxx8Rbxv3eSfbO-_29DipwpstrV2pDQ,3109
 natural_pdf/describe/summary.py,sha256=cfT4ZQkeatCDAOwWPwhtEVXisNgk6E57fAXAnoRysSU,7645
 natural_pdf/elements/__init__.py,sha256=ICNikmLeIEuSYypz-KnkBn8xR1hR7rge4hsa1KLkyWY,42
-natural_pdf/elements/base.py,sha256=92ukTtRCQFsa5KvKflChCt4mt0ZGS4ecGYCQTNMO4zU,58907
-natural_pdf/elements/element_collection.py,sha256=idM_BUWEfbCJ5Sq0Ae_KfbVHy8TdkNfzs7iWkFe_j2I,130707
+natural_pdf/elements/base.py,sha256=iJHEejlYu8RNvlKYK2UHAnAlz6tXkiEaGnG2xYtVnuU,59635
+natural_pdf/elements/element_collection.py,sha256=CVfnprzKTLeGSpvhGL2ZQVzZ8veSoFtCBlSSGDmX5lY,136594
 natural_pdf/elements/image.py,sha256=zu-P2Y8fRoEXf6IeZU0EYRWsgZ6I_a5vy1FA3VXTGkQ,1424
 natural_pdf/elements/line.py,sha256=TFn7KXjPT_jUQyQyabU0F7XYU4dC-qadwodJMZF4DCU,3844
-natural_pdf/elements/rect.py,sha256=0lNkVkPkvbRbrFED856RXoUcTcDkeeOIs5xldKGAQT8,3324
-natural_pdf/elements/region.py,sha256=HF6KzeuudO9upVLIrPsp3omcziLcILE3nnzl1a-LvK0,165400
-natural_pdf/elements/text.py,sha256=829uSJv9E-8cC6T6iR_Va7Xtv54pJoyRN78fq4NN1d4,20687
+natural_pdf/elements/rect.py,sha256=kmUmhwnihd-aTweAO-LsngRDo5Iqmx7lcSa8ZBlE_2E,4544
+natural_pdf/elements/region.py,sha256=DM8o0ptm86B2ouOqDgUK_av9cCN6G5iIlJ1VC3KfVWk,167379
+natural_pdf/elements/text.py,sha256=dOiss-cSBYnK9j7KqmqmvJcCidBcIhckLJCW8lVz2es,21210
 natural_pdf/export/mixin.py,sha256=L1q3MIEFWuvie4j4_EmW7GT3NerbZ1as0XMUoqTS7gM,5083
 natural_pdf/exporters/__init__.py,sha256=QffoARekR6WzXEd05oxOytly4qPdBizuIF-SUkeFpig,643
 natural_pdf/exporters/base.py,sha256=379sioW_hbkGb21sEVuJhbkkDO5MFsFtTUNO5TgG2YU,2101
@@ -85,7 +85,7 @@ natural_pdf/search/search_options.py,sha256=sq_e8_jSROicD94b_xtDtLnjEr_Zsy4icjzP
 natural_pdf/search/search_service_protocol.py,sha256=u8pbuWP96fnQEe6mnreY9DrdiDAHP6ZCY7phvSbFlP8,6697
 natural_pdf/search/searchable_mixin.py,sha256=hqQ_AuID5eTGRCtKYdFLZ1zF35y73uk3x1M1VW9Il8U,23514
 natural_pdf/selectors/__init__.py,sha256=oZGeqSv53EqmIZOhcnawuaGGlRg1h79vArXuZCWKm4A,123
-natural_pdf/selectors/parser.py,sha256=yV5Eb0VyNZocoYIXi7SMKsf8o66vrGNb-MeT27aEj-M,38977
+natural_pdf/selectors/parser.py,sha256=WOoJlCxyz6F50uWJEBWSucANWMvihroezMBZMPCZxzE,40388
 natural_pdf/tables/__init__.py,sha256=sCvCGbGsL6BiqlNxAYfVv003bIDLI11FmjHhaWfcU6w,104
 natural_pdf/tables/result.py,sha256=-8ctA-jCJYSHtlfAoqTvhUwO5zSP2BQxxetAjqEsNyg,8665
 natural_pdf/templates/__init__.py,sha256=jYBxzfi73vew0f6yhIh1MlRxw4F_TVN2hKQR0YXOFe0,20
@@ -101,38 +101,39 @@ natural_pdf/utils/packaging.py,sha256=TM0jafwS5yVbTGC-RMi4TyWunf9cUUo9h5J6rMzkT-
 natural_pdf/utils/reading_order.py,sha256=u7XyVZdKMPMK0CL1C7xFogKnZ92b0JKT068KFjQWe18,7437
 natural_pdf/utils/text_extraction.py,sha256=CCwPTmMoTgtQt2P00X_ADIf6ZGNfxvjCO9FO0_HqG40,13900
 natural_pdf/utils/visualization.py,sha256=zhZEHgYnZFuX7YxTHXF8Y3D97uHp2beTKMaC-JkCFwk,22364
-natural_pdf/vision/__init__.py,sha256=RymMY-3WLQBlOZ4Dx4MmL9UH6I65hNjkwUJ7ymO5JfM,287
-natural_pdf/vision/mixin.py,sha256=OJwBABr74TWxP5seTKUmGj5zE9mWsBP_UKWU-Pr8V9A,8720
-natural_pdf/vision/results.py,sha256=F2zXG3MVZIpOUvPkJHotOq6-9rFz68BaO_8pnSndlOs,5119
-natural_pdf/vision/similarity.py,sha256=YH8legN-t9uf1b_XULi4JLNDaRfPNKQwU1FZ4Qu08jY,11740
+natural_pdf/vision/__init__.py,sha256=TkoQtdODlh0n_99dsjLIWKE9dgK0m4jfrui_cQ3gTwU,221
+natural_pdf/vision/mixin.py,sha256=wlsX42cFUnUepZHsEfKBqXiDEPUwBG6-KN2Cx5qz_lw,10812
+natural_pdf/vision/results.py,sha256=_NBRCKtDd1M3sWK7zHSym7-jpQqW4kR_iFFL4PvnBNo,6649
+natural_pdf/vision/similarity.py,sha256=HWmXDBNLSOlRWH-_1K3FVR7tSsRuMFqXZwrVhhg2ZzU,17925
+natural_pdf/vision/template_matching.py,sha256=91XQt5tp-vmcMX_4b2Bz-YwIAlb-hc8E5ih_qAHQuCk,7145
 natural_pdf/widgets/__init__.py,sha256=QTVaUmsw__FCweFYZebwPssQxxUFUMd0wpm_cUbGZJY,181
 natural_pdf/widgets/viewer.py,sha256=KW3JogdR2TMg2ECUMYp8hwd060hfg8EsYBWxb5IEzBY,24942
-natural_pdf-0.2.12.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
+natural_pdf-0.2.15.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
 optimization/memory_comparison.py,sha256=0i_foFSRmppj-fY069qjwH36s_zkx-1L2ASAAlepWzA,6541
 optimization/pdf_analyzer.py,sha256=HjrmTgu2qchxPeDckc5kjgxppGwd40UESrYS9Myj7pY,19352
 optimization/performance_analysis.py,sha256=JBXnR9hc7Ix7YCnt3EJPSpsyqIUgKsc7GEffQ_TDCBk,13033
 optimization/test_cleanup_methods.py,sha256=PmLOL4MRgvV0j_DW9W1TS8MsGGgu57QCuq6_5y7zK3s,6209
 optimization/test_memory_fix.py,sha256=A3knK74fNhvHknDbLhbTmA276x1ifl-3ivJ_7BhVSTI,6170
-temp/debug_cell_extraction.py,sha256=nE0Z470P40v8xZfWO1V3qgNaejs_pernEQaUOFeOJ1U,1527
-temp/debug_exclusion_overlap.py,sha256=RptJXwqBXy5gsvMF037KEx1o2QgjwEDkMB6TD5aJdqA,1644
-temp/debug_exclusions_guides.py,sha256=s8siep9te1KRJ2j0vH1tvDQnBlz7PKbHeCiYMrZL8jE,2096
-temp/debug_extra_guide.py,sha256=95Tim-YnmAR4kICw2XDKVDvlW5WsjK_51cv5-EV11rc,1236
-temp/debug_outer_boundaries.py,sha256=uJUJwojTxOU4VtbGUouuhV65IYzS6NDIVKxnS7o64nU,1456
-temp/debug_st_search.py,sha256=F4c_mUVi_d5AKaKIpQ0AnW1amDqAwALoQQj7wZj--J0,1021
 temp/fix_page_exclusions.py,sha256=YIj62zF38TdoBARAuSIvEbetl_JfXG-mp4v9p355qmo,1358
+temp/test_draw_guides.py,sha256=_eSSBElGHQkd2QD_KA_Okw70v0dlY5m-1-C5SQwKAJw,642
+temp/test_draw_guides_interactive.py,sha256=FsH-2ZQGsGx_8QfVCWUAkLbOcJz-VfiwROzQD4AD7kQ,926
 temp/test_exclusion_with_debug.py,sha256=CScxHvb43KrB5dzXuTOhuzjcBXZBdfYB5ygiKkEW26g,1393
 temp/test_find_exclusions_fix.py,sha256=1l5aEqnElcl3kiykdtmJFlVxQ1xMKGm1UckGYEQg--c,2103
 temp/test_find_exclusions_fix_no_recursion.py,sha256=qZspTBwxunRM93N_-fZ2fR5Lodj0ArQX3h10HlTXhfc,3592
 temp/test_fix_real_pdf.py,sha256=uuylxmpeAEbIix9wjl0Gri1sZlN61dBWTq6ZCyfvzF8,1454
 temp/test_fix_working.py,sha256=-Ryre1rXYA2EG_lmPZGYEGi8yz0slhHEXPJMYexZW84,1750
 temp/test_fixed_pdf_exclusions.py,sha256=Q5zxooKDvtTXo-dDsx3nsQw1ZVHX3TW47iZ_dXpFdrY,2168
+temp/test_guide_draw_notebook.py,sha256=9yYRV5mfmVHiL1lnwNj-vksw45d1oWbAZpDGA7yZf-M,1583
 temp/test_horizontal_top_bottom.py,sha256=Mb3tjt9Z3wOTpzFOgK7i0K-j-_ynNh4vDu2x1L3nu-s,2163
+temp/test_inline_js.py,sha256=xuQH8VQn7L4sogv6wd_Rwudx5p_Lt6we1h7U1LPTH-g,646
 temp/test_marker_order.py,sha256=TFZkMxRiNoZGVcdDivYnkIDNvwHaiyKUdYoy2rTTIiI,1417
 temp/test_original_exclusions_now_work.py,sha256=G6LmaF-P9Qhj0j4lT_4ncfCddllfP6L8F_x2prUBr9w,1904
 temp/test_pdf_exclusions_with_guides.py,sha256=QaMl0frgKC8kCPQ2BUI8kqyvqsIjQPXKV_St1rK3zxg,2754
 temp/test_region_exclusions_detailed.py,sha256=EftdW3JY3JH_LX5QlWKt-4drM-joPggK2fKUZRXVTMA,814
 temp/test_stripes_real_pdf.py,sha256=FIvDoJrnuioOMw1A0aTCCfZLeg99lusfe0Fb0MiqnhQ,2618
 temp/test_vertical_stripes.py,sha256=Yf3TJfb_faqAFzlgb7i5u6dDHjF4UMSHIGM99vangRk,1877
+temp/test_widget_functionality.py,sha256=jsEGHYK1dWWa8uEcfGRRj1ReHRMzNoIaMZU4d-o-Djs,2448
+temp/test_widget_simple.py,sha256=Vy_DKgPhPhUQ8nKw_KnhGTpwtmh5EEic0avEyW9hbOQ,1398
 tools/bad_pdf_eval/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 tools/bad_pdf_eval/analyser.py,sha256=oqSTo3NLyignp_XdCO9_SRCUUXMU8lfgDavKYZYNxws,13690
 tools/bad_pdf_eval/collate_summaries.py,sha256=L_YsdiqmwGIHYWTVJqo6gyazyn3GIQgpfGGKk8uwckk,5159
@@ -144,8 +145,8 @@ tools/bad_pdf_eval/llm_enrich.py,sha256=mCh4KGi1HmIkzGjj5rrHz1Osd7sEX1IZ_FW08H1t
 tools/bad_pdf_eval/llm_enrich_with_retry.py,sha256=XUtPF1hUvqd3frDXT0wDTXoonuAivhjM5vgFdZ-tm0A,9373
 tools/bad_pdf_eval/reporter.py,sha256=e1g__mkSB4q02p3mGWOwMhvFs7F2HJosNBxup0-LkyU,400
 tools/bad_pdf_eval/utils.py,sha256=hR95XQ7qf7Cu6BdyX0L7ggGVx-ah5sK0jHWblTJUUic,4896
-natural_pdf-0.2.12.dist-info/METADATA,sha256=jRNM0JxYvPDuqzD63earjbaUwQgXCjPYPLC5pLl49Uk,6960
-natural_pdf-0.2.12.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-natural_pdf-0.2.12.dist-info/entry_points.txt,sha256=1R_KMv7g60UBBpRqGfw7bppsMNGdayR-iJlb9ohEk_8,81
-natural_pdf-0.2.12.dist-info/top_level.txt,sha256=ZDKhxE_tg508o9BpagsjCGcI8GY4cF_8bg0e0IaLsPI,41
-natural_pdf-0.2.12.dist-info/RECORD,,
+natural_pdf-0.2.15.dist-info/METADATA,sha256=TfUHRd_THs6bk7Iwl8UdBjZfnWEXKg_tBiVn_Q9tv28,6960
+natural_pdf-0.2.15.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+natural_pdf-0.2.15.dist-info/entry_points.txt,sha256=1R_KMv7g60UBBpRqGfw7bppsMNGdayR-iJlb9ohEk_8,81
+natural_pdf-0.2.15.dist-info/top_level.txt,sha256=ZDKhxE_tg508o9BpagsjCGcI8GY4cF_8bg0e0IaLsPI,41
+natural_pdf-0.2.15.dist-info/RECORD,,

temp/test_draw_guides.py

@@ -0,0 +1,25 @@
+"""Example usage of the interactive guide drawing feature"""
+
+# In a Jupyter notebook:
+from natural_pdf import NaturalPDF
+
+# Load a PDF
+pdf = NaturalPDF.from_file("your_pdf.pdf")
+page = pdf[0]
+
+# Create guides
+guides = page.guides()
+
+# Detect some initial guides (optional)
+guides.vertical.from_lines(n=5)
+guides.horizontal.from_lines(n=5)
+
+# Open interactive editor for vertical guides
+guides.vertical.draw()
+
+# Open interactive editor for horizontal guides
+guides.horizontal.draw(width=600)  # Smaller widget
+
+# After editing, the guides are automatically updated
+# You can now use them to extract tables:
+table = page.extract_table(guides)

temp/test_draw_guides_interactive.py

@@ -0,0 +1,30 @@
+"""Test the interactive guide drawing functionality"""
+
+from natural_pdf.core.pdf import PDF
+from natural_pdf.analyzers.guides import Guides
+
+# Load a sample PDF
+pdf = PDF("tests/sample_pdfs/simple_table.pdf")
+page = pdf.pages[0]
+
+# Create guides
+guides = Guides(page)
+
+# Add some initial guides for testing
+guides.vertical.from_content()
+guides.horizontal.from_lines(n=5)
+
+print("Initial vertical guides:", list(guides.vertical))
+print("Initial horizontal guides:", list(guides.horizontal))
+
+# This would open the interactive widget in Jupyter
+# guides.vertical.draw()
+
+# For non-Jupyter testing, we can check the method exists
+assert hasattr(guides.vertical, 'draw')
+assert callable(guides.vertical.draw)
+
+print("\nSuccess! The draw() method is available on GuidesList objects.")
+print("To use it interactively, run this in a Jupyter notebook:")
+print("  guides.vertical.draw()")
+print("  guides.horizontal.draw(width=600)")

temp/test_guide_draw_notebook.py

@@ -0,0 +1,47 @@
+"""Test script to verify the draw() method works"""
+
+import sys
+sys.path.insert(0, '.')
+
+from natural_pdf.analyzers.guides import GuidesList, Guides
+
+# Create a mock context for testing
+class MockContext:
+    def __init__(self):
+        self.width = 600
+        self.height = 800
+        
+    def render(self, resolution=150):
+        # Create a simple test image
+        from PIL import Image
+        img = Image.new('RGB', (int(self.width * resolution/72), int(self.height * resolution/72)), 'white')
+        return img
+
+# Test that the draw method exists
+mock_context = MockContext()
+guides = Guides(mock_context)
+
+# Add some test guides
+guides.vertical.data = [100, 200, 300, 400, 500]
+guides.horizontal.data = [150, 350, 550, 750]
+
+print("Initial vertical guides:", list(guides.vertical))
+print("Initial horizontal guides:", list(guides.horizontal))
+
+# Check that draw method exists
+assert hasattr(guides.vertical, 'draw')
+assert callable(guides.vertical.draw)
+assert hasattr(guides.horizontal, 'draw')
+assert callable(guides.horizontal.draw)
+
+print("\nSuccess! The draw() method is available.")
+print("\nIn a Jupyter notebook, you would use:")
+print("  guides.vertical.draw()      # Interactive vertical guide editor")
+print("  guides.horizontal.draw()    # Interactive horizontal guide editor")
+print("\nFeatures:")
+print("  - Click to add new guides")
+print("  - Click existing guides to select them")
+print("  - Drag to move guides")
+print("  - Delete key to remove selected guide")
+print("  - Arrow keys to fine-tune position")
+print("  - Enter to apply, Escape to cancel")

temp/test_inline_js.py

@@ -0,0 +1,22 @@
+"""Test inline JavaScript in HTML widget"""
+
+import ipywidgets as widgets
+from IPython.display import display
+
+# Create an HTML widget with inline JavaScript
+html_content = '''
+<div id="test-div">Click me!</div>
+<script type="text/javascript">
+document.getElementById('test-div').addEventListener('click', function() {
+    alert('Clicked!');
+    this.innerHTML = 'Clicked at ' + new Date().toLocaleTimeString();
+});
+console.log('JavaScript is running!');
+</script>
+'''
+
+# Display using widgets.HTML
+html_widget = widgets.HTML(value=html_content)
+display(html_widget)
+
+print("If you see 'Click me!' above and can click it, JavaScript is working.")

temp/test_widget_functionality.py

@@ -0,0 +1,68 @@
+#!/usr/bin/env python
+"""Test the guide widget functionality"""
+
+import sys
+import os
+
+# Add parent directory to path
+sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
+
+# Test importing and basic functionality
+try:
+    from natural_pdf.analyzers.guides import InteractiveGuideWidget, GuidesList, _GUIDE_WIDGET_AVAILABLE
+    print("✓ Successfully imported InteractiveGuideWidget")
+    
+    if _GUIDE_WIDGET_AVAILABLE:
+        print("✓ ipywidgets is available")
+        
+        # Create a mock GuidesList for testing
+        class MockPage:
+            def __init__(self):
+                self.bbox = (0, 0, 595, 842)  # A4 page size in points
+                
+            def render(self, resolution=150):
+                # Mock render method
+                from PIL import Image
+                width = int(595 * resolution / 72)
+                height = int(842 * resolution / 72)
+                return Image.new('RGB', (width, height), color='white')
+                
+        class MockGuides:
+            def __init__(self):
+                self.context = MockPage()
+                
+        class MockGuidesList:
+            def __init__(self):
+                self.data = [100, 200, 300]
+                self._axis = 'vertical'
+                self._parent = MockGuides()
+        
+        # Test creating the widget
+        mock_guides = MockGuidesList()
+        try:
+            widget = InteractiveGuideWidget(mock_guides)
+            print("✓ Successfully created InteractiveGuideWidget instance")
+            print(f"  - Widget ID: {widget.widget_id}")
+            print(f"  - Widget base classes: {InteractiveGuideWidget.__bases__}")
+            
+            # Check if the widget has the expected methods
+            expected_methods = ['_generate_content', 'update_guides']
+            for method in expected_methods:
+                if hasattr(widget, method):
+                    print(f"  - Has method: {method}")
+                else:
+                    print(f"  - Missing method: {method}")
+                    
+        except Exception as e:
+            print(f"✗ Error creating widget: {e}")
+            
+    else:
+        print("⚠ ipywidgets not available - widget functionality disabled")
+        
+except ImportError as e:
+    print(f"✗ Import error: {e}")
+    
+except Exception as e:
+    print(f"✗ Unexpected error: {e}")
+    
+print("\nWidget implementation test complete!")

temp/test_widget_simple.py

@@ -0,0 +1,41 @@
+#!/usr/bin/env python
+"""Simple test for the guide widget"""
+
+import sys
+import os
+
+# Add parent directory to path
+sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), '..')))
+
+# Test importing the module
+try:
+    from natural_pdf.analyzers.guides import InteractiveGuideWidget, _GUIDE_WIDGET_AVAILABLE
+    print(f"✓ Module imported successfully")
+    print(f"✓ Widget available: {_GUIDE_WIDGET_AVAILABLE}")
+    
+    if _GUIDE_WIDGET_AVAILABLE:
+        print("✓ ipywidgets is installed and InteractiveGuideWidget is available")
+    else:
+        print("✗ ipywidgets is not installed")
+        
+except ImportError as e:
+    print(f"✗ Import error: {e}")
+    sys.exit(1)
+
+# Check if we can create the widget class
+if _GUIDE_WIDGET_AVAILABLE:
+    try:
+        # We can't actually instantiate it without a GuidesList, but we can check the class exists
+        print(f"✓ InteractiveGuideWidget class: {InteractiveGuideWidget}")
+        print(f"✓ Widget base classes: {InteractiveGuideWidget.__bases__}")
+        
+        # Check methods
+        methods = [m for m in dir(InteractiveGuideWidget) if not m.startswith('_')]
+        print(f"✓ Public methods: {methods}")
+        
+    except Exception as e:
+        print(f"✗ Error checking widget class: {e}")
+else:
+    print("⚠ Skipping widget checks as ipywidgets is not available")
+
+print("\nAll checks passed!")

temp/debug_cell_extraction.py

@@ -1,42 +0,0 @@
-"""Debug cell text extraction with exclusions"""
-from natural_pdf import PDF
-from natural_pdf.analyzers.guides import Guides
-
-pdf = PDF("pdfs/m27.pdf")
-page = pdf.pages[0]
-
-# Add exclusions
-pdf.add_exclusion(lambda page: page.find(text="PREMISE").above(), label="header")
-
-# Check exclusions are registered
-print("Exclusions on page:")
-exclusions = page._get_exclusion_regions(debug=True)
-
-# Create guides and build grid
-headers = page.find(text="NUMBER").right(include_source=True).expand(top=3, bottom=3).find_all('text')
-guides = Guides(page)
-guides.vertical.from_content(headers, align='left')
-guides.horizontal.from_stripes()
-
-# Build grid and get cells
-grid_result = guides.build_grid(include_outer_boundaries=True)
-cells = grid_result["regions"]["cells"]
-
-print(f"\nTotal cells: {len(cells)}")
-
-# Check first row cells (these should be in excluded area)
-first_row_cells = [c for c in cells if c.bbox[1] < 90]  # y < 90
-print(f"\nFirst row cells: {len(first_row_cells)}")
-
-for i, cell in enumerate(first_row_cells[:3]):
-    print(f"\nCell {i}:")
-    print(f"  Bbox: {cell.bbox}")
-    print(f"  Raw text: {repr(cell.extract_text(apply_exclusions=False))}")
-    print(f"  With exclusions: {repr(cell.extract_text(apply_exclusions=True))}")
-
-# Now test the full table extraction
-print("\n\nFull table extraction:")
-result = guides.extract_table(include_outer_boundaries=True, apply_exclusions=True, header=False)
-df = result.to_df()
-print("\nFirst row of dataframe:")
-print(df.iloc[0].to_dict() if not df.empty else "Empty")

temp/debug_exclusion_overlap.py

@@ -1,43 +0,0 @@
-"""Debug how exclusions work with overlapping regions"""
-from natural_pdf import PDF
-from natural_pdf.analyzers.guides import Guides
-
-pdf = PDF("pdfs/m27.pdf")
-page = pdf.pages[0]
-
-# Add exclusion
-pdf.add_exclusion(lambda page: page.find(text="PREMISE").above(), label="header")
-
-# Get the exclusion region
-exclusions = page._get_exclusion_regions()
-excl_region = exclusions[0]
-print(f"Exclusion region: {excl_region.bbox}")
-print(f"Exclusion bottom: {excl_region.bbox[3]}")
-
-# Create a test cell that overlaps the exclusion
-# Cell 1 from before: (32.06, 0.5, 73.18288, 79.53999999999996)
-test_cell = page.region(32.06, 0.5, 73.18288, 79.53999999999996)
-
-print(f"\nTest cell: {test_cell.bbox}")
-print(f"Cell overlaps exclusion: top={test_cell.bbox[1]} < excl_bottom={excl_region.bbox[3]}")
-
-# Extract text from different y-ranges
-print("\nText in different parts of the cell:")
-
-# Part above exclusion line (should be empty)
-upper_part = page.region(32.06, 0.5, 73.18288, 59.12)
-print(f"Upper part (0.5 to 59.12): '{upper_part.extract_text(apply_exclusions=True)}'")
-
-# Part below exclusion line (should have text)  
-lower_part = page.region(32.06, 59.12, 73.18288, 79.54)
-print(f"Lower part (59.12 to 79.54): '{lower_part.extract_text()}'")
-
-# The whole cell
-print(f"Whole cell with exclusions: '{test_cell.extract_text(apply_exclusions=True)}'")
-print(f"Whole cell without exclusions: '{test_cell.extract_text(apply_exclusions=False)}'")
-
-# Check what text elements are in this region
-print("\nText elements in cell:")
-cell_texts = test_cell.find_all('text')
-for t in cell_texts[:5]:
-    print(f"  '{t.text}' at y={t.top:.2f}-{t.bottom:.2f}")