can table exacted without lines

[tujinshu]

the table(without lines) in pdf can't be exacted (page 4)

is pdfplumber able to handle this ?
tencent.pdf

[jsvine]

You might try cropping to the part of the PDF with the table, and then using the "horizontal_strategy": "text", "vertical_strategy": "text" table-extraction settings. If that doesn't work, you could try programmatically identifying the vertical and horizontal divisions (making use of, for instance, the positions of those short horizontal lines/rects), and then using "explicit_horizontal_lines": [...], "explicit_vertical_lines": [...].

[tujinshu]

thank you for your reply,i will have a try！

[tujinshu]

i use text_strategy to handle this tencent.pdf

p0 = pdf.pages[3]
p0 = p0.filter(keep_visible_lines)
ts = {           
            #Tables without lines and borders
            "vertical_strategy": "text",
            "horizontal_strategy": "text",
            "explicit_vertical_lines": [],
            "explicit_horizontal_lines": [],
            "snap_tolerance": 3,
            "snap_x_tolerance": 3,
            "snap_y_tolerance": 3,
            "join_tolerance": 3,
            "join_x_tolerance": 3,
            "join_y_tolerance": 3,
            "edge_min_length": 3,
            "min_words_vertical": 3,
            "min_words_horizontal": 1,
            #I get error with this line active, because I'm using find_tables
            "text_tolerance": 3,
            "text_x_tolerance": 3,
            "text_y_tolerance": 3,
            "intersection_tolerance": 3,
            "intersection_x_tolerance": 3,
            "intersection_y_tolerance": 3,
        }

im = p0.to_image()
im.debug_tablefinder(ts)

but it generate two row with origin one row , for example(in green cicyle) :

is there any way to optimize this ?

[cmdlineluser]

Update: As the data seems well formed, perhaps it's simpler to crop and just extract the text lines?

hlines = {}
for edge in page.horizontal_edges:
   if edge['object_type'] == 'line':
      left, right = edge['x0'], edge['x1']
      hlines[left, right] = edge
      
left, right = min(hlines)[0], max(hlines)[1]
crop = page.crop((left, 0, right, page.height))

[ line['text'].split() for line in crop.extract_text_lines() ]

[['財務概要'],
 ['截至十二月三十一日止年度'],
 ['二零一八年', '二零一九年', '二零二零年', '二零二一年', '二零二二年'],
 ['人民幣百萬元', '人民幣百萬元', '人民幣百萬元', '人民幣百萬元', '人民幣百萬元'],
 ['312,694', '377,289', '482,064', '560,118', '554,552'],
 ['142,120', '167,533', '221,532', '245,944', '238,746'],
 ['94,466', '109,400', '180,022', '248,062', '210,225'],
 ['79,984', '95,888', '160,125', '227,810', '188,709'],
 ['78,719', '93,310', '159,847', '224,822', '188,243'],
 ['67,760', '119,901', '281,173', '200,390', '59,564'],
 ['66,339', '116,670', '277,834', '200,323', '60,699'],
 ['92,481', '114,601', '149,404', '159,539', '153,538'],
 ['77,469', '94,351', '122,742', '123,788', '115,649'],
 ['於十二月三十一日'],
 ['二零一八年', '二零一九年', '二零二零年', '二零二一年', '二零二二年'],
 ['人民幣百萬元', '人民幣百萬元', '人民幣百萬元', '人民幣百萬元', '人民幣百萬元'],
 ['506,441', '700,018', '1,015,778', '1,127,552', '1,012,142'],
 ['217,080', '253,968', '317,647', '484,812', '565,989'],
 ['723,521', '953,986', '1,333,425', '1,612,364', '1,578,131'],
 ['323,510', '432,706', '703,984', '806,299', '721,391'],
 ['32,697', '56,118', '74,059', '70,394', '61,469'],
 ['356,207', '488,824', '778,043', '876,693', '782,860'],
 ['164,879', '225,006', '286,303', '332,573', '361,067'],
 ['202,435', '240,156', '269,079', '403,098', '434,204'],
 ['367,314', '465,162', '555,382', '735,671', '795,271'],
 ['723,521', '953,986', '1,333,425', '1,612,364', '1,578,131'],
 ['二零二二年年報', '3']]

---

Maybe this is helpful:

You could target the horizontal lines as @jsvine suggested:

im = page.to_image(300)

hlines = {}
for edge in page.horizontal_edges:
   if edge['object_type'] == 'line':
      left, right = edge['x0'], edge['x1']
      hlines[left, right] = edge
      
# im.reset().draw_lines(hlines.values(), stroke_width=10).save('lines.png')

You could then extract the words that fall within the width of those lines:

words = {}
for word in page.extract_words():
    text, x0, x1, top, bottom = (
        word['text'], word['x0'], word['x1'], word['top'], word['bottom']
    )
    for (left, right) in hlines:
        if  (((x0 >= left)   or (abs(left  - x0) < 0.5))
        and  ((x1 <= right)) or (abs(right - x1) < 0.5)):
            words[text, x0, x1, top, bottom] = word
            break
            
# im.reset().draw_rects(words.values()).save('words.png')

You could then group the words into "rows" based on their "top" position:

rows = []
row  = []
for word in words.values():
    if len(rows) == 0 and len(row) == 0:
        row.append(word)
        continue
    if len(row) == 0:
        row.append(word)
        continue
    if abs(word['top'] - row[-1]['top']) < 1:
        row.append(word)
    else:
        rows.append(row)
        row = []
        row.append(word)
rows.append(row)
        
data = [[col['text'] for col in row] for row in rows]

[['二零一八年', '二零一九年', '二零二零年', '二零二一年', '二零二二年'],
 ['人民幣百萬元', '人民幣百萬元', '人民幣百萬元', '人民幣百萬元', '人民幣百萬元'],
 ['312,694', '377,289', '482,064', '560,118', '554,552'],
 ['142,120', '167,533', '221,532', '245,944', '238,746'],
 ['94,466', '109,400', '180,022', '248,062', '210,225'],
 ['79,984', '95,888', '160,125', '227,810', '188,709'],
 ['78,719', '93,310', '159,847', '224,822', '188,243'],
 ['67,760', '119,901', '281,173', '200,390', '59,564'],
 ['66,339', '116,670', '277,834', '200,323', '60,699'],
 ['92,481', '114,601', '149,404', '159,539', '153,538'],
 ['77,469', '94,351', '122,742', '123,788', '115,649'],
 ['二零一八年', '二零一九年', '二零二零年', '二零二一年', '二零二二年'],
 ['人民幣百萬元', '人民幣百萬元', '人民幣百萬元', '人民幣百萬元', '人民幣百萬元'],
 ['506,441', '700,018', '1,015,778', '1,127,552', '1,012,142'],
 ['217,080', '253,968', '317,647', '484,812', '565,989'],
 ['723,521', '953,986', '1,333,425', '1,612,364', '1,578,131'],
 ['323,510', '432,706', '703,984', '806,299', '721,391'],
 ['32,697', '56,118', '74,059', '70,394', '61,469'],
 ['356,207', '488,824', '778,043', '876,693', '782,860'],
 ['164,879', '225,006', '286,303', '332,573', '361,067'],
 ['202,435', '240,156', '269,079', '403,098', '434,204'],
 ['367,314', '465,162', '555,382', '735,671', '795,271'],
 ['723,521', '953,986', '1,333,425', '1,612,364', '1,578,131']]

[tujinshu]

thank you for your reply，

is divide into two rows, 312,694 and ______ ，is there any parameter can combine them

[cmdlineluser]

I'm not sure I understand correctly, these are the rows I get:

│ 312,694      ┆ 377,289      ┆ 482,064      ┆ 560,118      ┆ 554,552      │
│ 142,120      ┆ 167,533      ┆ 221,532      ┆ 245,944      ┆ 238,746      │
│ 94,466       ┆ 109,400      ┆ 180,022      ┆ 248,062      ┆ 210,225      │

Where does _____ come into play?