Returns a specification to start this module under a supervisor.

See Supervisor.

Extracts metadata from a PDF file info trailers. Typically includes "CreationDate", "ModDate", "Producer", et cetera.

Examples

iex> PdfExtractor.extract_metadata("priv/fixtures/frankenstein-mary-shelley.pdf")
{:ok, %{"CreationDate" => "D:20260403141150", "Creator" => "PDFium", "Producer" => "PDFium"}}

Extracts metadata from PDF binary data. Similar to extract_metadata/1 but works with PDF data in memory instead of files.

Examples

iex> content = File.read!("priv/fixtures/frankenstein-mary-shelley.pdf")
...> PdfExtractor.extract_metadata_from_binary(content)
{:ok, %{"CreationDate" => "D:20260403141150", "Creator" => "PDFium", "Producer" => "PDFium"}}

Extracts the size and rotation of PDF pages.

Each page maps to its :width and :height in PDF points and its :rotation in degrees, one of 0, 90, 180 or 270. The sizes are the frame used by the areas of the other functions, so read them before choosing a bounding box. When all pages are requested, the size of the map is the number of pages of the document.

Pages are an integer or a list of page numbers as in extract_text/3. Areas do not apply here. Accepts the :timeout, :name and :password options.

Examples

Get the size of every page.

iex> PdfExtractor.extract_page_info("priv/fixtures/fatura.pdf")
{:ok,
 %{
   0 => %{width: 612.0, height: 792.0, rotation: 0},
   1 => %{width: 595.2756, height: 841.8898, rotation: 0}
 }}

Count the pages of a document.

iex> {:ok, pages} =
...>   PdfExtractor.extract_page_info("priv/fixtures/frankenstein-mary-shelley.pdf")
...>
...> map_size(pages)
17

Extracts the size and rotation of pages from PDF binary data. See extract_page_info/3 for details.

Examples

iex> content = File.read!("priv/fixtures/fatura.pdf")
...> PdfExtractor.extract_page_info_from_binary(content, 1)
{:ok, %{1 => %{width: 595.2756, height: 841.8898, rotation: 0}}}

Extracts tables from PDF pages together with their position on the page.

Each page maps to the list of tables found on it. A table is a map with its :rows, each one a list of cells read left to right, and the edges of its bounding box, :left, :right, :top and :bottom, in PDF points measured from the top left corner of the page. A cell is nil when the table has a cell there but no text could be assigned to it. The bounding box can be used as an area with the other functions, for example to read the text right below a table.

See extract_text/3 for details on how to specify pages, areas, and options.

Options

Besides :timeout, :name and :password, this function accepts how the edges of the tables are found:

• :vertical_strategy - how to find the columns (default: :lines)
• :horizontal_strategy - how to find the rows (default: :lines)

Each strategy is one of:

• :lines - use the lines drawn on the page, including the sides of rectangles, as cell borders
• :lines_strict - like :lines but ignore the sides of rectangles
• :text - infer the borders from the alignment of the words, for tables drawn without lines

Examples

Extract the tables of every page.

iex> PdfExtractor.extract_tables("priv/fixtures/fatura.pdf")
{:ok,
 %{
   0 => [
     %{
       rows: [["Item", "Quantidade", "Avaliar", "Quantia"]],
       left: 29.25,
       top: 190.5,
       right: 582.75,
       bottom: 211.5
     }
   ],
   1 => []
 }}

Extract only the tables inside an area.

iex> PdfExtractor.extract_tables("priv/fixtures/fatura.pdf", %{0 => {0, 180, 595, 260}})
{:ok,
 %{
   0 => [
     %{
       rows: [["Item", "Quantidade", "Avaliar", "Quantia"]],
       left: 29.25,
       top: 190.5,
       right: 582.75,
       bottom: 211.5
     }
   ]
 }}

Find the columns from the alignment of the words instead of the drawn lines.

iex> PdfExtractor.extract_tables("priv/fixtures/fatura.pdf", 0,
...>   vertical_strategy: :text,
...>   horizontal_strategy: :lines
...> )
{:ok,
 %{
   0 => [
     %{
       rows: [
         [nil, nil, "devedor:", "", "1 52", "5,59 €"],
         [nil, nil, "", "", "", ""],
         ["I", "tem Quantidad", "e", "Avaliar", "", "Quantia"]
       ],
       left: 44.25,
       top: 126.75,
       right: 568.0695,
       bottom: 211.5
     }
   ]
 }}

Extracts tables from PDF binary data. See extract_tables/3 for the shape of each table and its options, and extract_text/3 for details on how to specify pages and areas.

Examples

iex> content = File.read!("priv/fixtures/fatura.pdf")
...> PdfExtractor.extract_tables_from_binary(content, 0)
{:ok,
 %{
   0 => [
     %{
       rows: [["Item", "Quantidade", "Avaliar", "Quantia"]],
       left: 29.25,
       top: 190.5,
       right: 582.75,
       bottom: 211.5
     }
   ]
 }}

Extracts text from PDF pages.

It supports extracting from single pages, multiple pages, and specific areas within pages.

Page Numbers

• Integer: Extract from single page (e.g., 0 for first page)
• List: Extract from multiple pages (e.g., [0, 1, 2])
• Empty list []: Extract from all pages (default)

Areas Format

Areas are specified as a map where keys are page numbers and values are bounding boxes:

• Single area: %{0 => {x0, y0, x1, y1}}
• Multiple areas: %{0 => [{x0, y0, x1, y1}, {x2, y2, x3, y3}]}
• Mixed: %{0 => {x0, y0, x1, y1}, 1 => [{x2, y2, x3, y3}, {x4, y4, x5, y5}]}

Options

All extraction functions accept these options:

• :timeout - timeout in milliseconds for the GenServer call (default: #{@default_timeout})
• :name - name or PID of the server to call (default: PdfExtractor)
• :password - the password of an encrypted PDF. Without it, or with a wrong one, extraction returns an error (default: nil)

extract_text/3 and extract_text_from_binary/3 also accept:

• :layout - when true, keeps the position of the text on the page by padding it with spaces and blank lines, so columns stay aligned. Useful for statements and tables that the default extraction flattens into single lines (default: false)

Examples

Extract text from all pages.

iex> PdfExtractor.extract_text("priv/fixtures/fatura.pdf")
{:ok,
 %{
   0 =>
     "Text Example Bill FATURA\n# 2025010002\nData: Jun 21, 2025\nProjeto de lei para:\nSaldo devedor: 1 525,59 €\nElixir Company\nItem Quantidade Avaliar Quantia\nTrabalho 1 1 500,00 € 1 500,00 €\nMais trabalho 1 25,59 € 25,59 €\nSubtotal: 1 525,59 €\nImposto (0%): 0,00 €\nTotal: 1 525,59 €",
   1 =>
     "✂\nReceipt Payment part Account / Payable to\nCH4431999123000889012\n✂\nMax Muster & Söhne\nAccount / Payable to\nCH4431999123000889012 Musterstrasse 123\nMax Muster & Söhne 8000 Seldwyla\nMusterstrasse 123\n8000 Seldwyla\nReference\n210000000003139471430009017\nReference\n210000000003139471430009017\nAdditional information\nBestellung vom 15.10.2020\nPayable by (name/address)\nSimon Muster\nPayable by (name/address)\nMusterstrasse 1\nCurrency Amount\nSimon Muster\n8000 Seldwyla\nCHF 1 949.75 Musterstrasse 1\n8000 Seldwyla\nCurrency Amount\nCHF 1 949.75\nAcceptance point"
 }}

Extract text from only some pages.

iex> PdfExtractor.extract_text("priv/fixtures/fatura.pdf", [0])
{:ok,
 %{
   0 =>
     "Text Example Bill FATURA\n# 2025010002\nData: Jun 21, 2025\nProjeto de lei para:\nSaldo devedor: 1 525,59 €\nElixir Company\nItem Quantidade Avaliar Quantia\nTrabalho 1 1 500,00 € 1 500,00 €\nMais trabalho 1 25,59 € 25,59 €\nSubtotal: 1 525,59 €\nImposto (0%): 0,00 €\nTotal: 1 525,59 €"
 }}

Extract only the titles in the book.

iex> PdfExtractor.extract_text("priv/fixtures/frankenstein-mary-shelley.pdf", %{
...>   1 => {0, 0, 612, 190},
...>   2 => {0, 0, 612, 190},
...>   8 => {0, 0, 612, 190},
...>   12 => {0, 0, 612, 190},
...>   13 => {0, 0, 612, 190}
...> })
{:ok,
 %{
   1 => "Frankenstein",
   2 => "Introduction",
   8 => "Preface",
   12 => "Frankenstein",
   13 => "Letter I"
 }}

Extract multiple areas from a single page.

iex> PdfExtractor.extract_text("priv/fixtures/frankenstein-mary-shelley.pdf", %{
...>   13 => [{0, 200, 600, 300}, {0, 400, 612, 440}]
...> })
{:ok,
 %{
   13 => [
     "To Mrs. Saville, England.\nSt. Petersburgh, Dec. 11th, 17 —.",
     "I am already far north of London; and as I walk in the streets of\nPetersburgh, I feel a cold northern breeze play upon my cheeks, which"
   ]
 }}

Extract with a custom timeout for large files.

iex> PdfExtractor.extract_text("priv/fixtures/frankenstein-mary-shelley.pdf", 11,
...>   timeout: 30_000
...> )
{:ok,
 %{
   11 =>
     "“Did I request thee, Maker, from my clay\nTo mould me Man, did I solicit thee\nFrom darkness to promote me?”\nP L , X, 743 – 45\nARADISE OST"
 }}

Open an encrypted file with its password.

iex> PdfExtractor.extract_text("priv/fixtures/fatura-encrypted.pdf", %{0 => {0, 0, 300, 200}},
...>   password: "secret"
...> )
{:ok, %{0 => "Text Example Bill\nProjeto de lei para:\nElixir Company"}}

Keep the layout of the page so the columns of a table stay aligned.

iex> PdfExtractor.extract_text("priv/fixtures/fatura.pdf", %{0 => {0, 220, 595, 250}})
{:ok, %{0 => "Trabalho 1 1 500,00 € 1 500,00 €\nMais trabalho 1 25,59 € 25,59 €"}}

iex> PdfExtractor.extract_text("priv/fixtures/fatura.pdf", %{0 => {0, 220, 595, 250}},
...>   layout: true
...> )
{:ok,
 %{
   0 =>
     "      Trabalho                                     1          1 500,00 € 1 500,00 €\n      Mais trabalho                                1            25,59 €   25,59 € "
 }}

Extracts text from PDF binary data. See extract_text/2 for details on how to specify pages, areas, and options.

This function allows you to extract text from PDF data that's already in memory, such as data downloaded from a URL or received via an API. This avoids the need to write the PDF to the filesystem.

Examples

Extract text from all pages.

iex> content = File.read!("priv/fixtures/fatura.pdf")
...> PdfExtractor.extract_text_from_binary(content)
{:ok,
 %{
   0 =>
     "Text Example Bill FATURA\n# 2025010002\nData: Jun 21, 2025\nProjeto de lei para:\nSaldo devedor: 1 525,59 €\nElixir Company\nItem Quantidade Avaliar Quantia\nTrabalho 1 1 500,00 € 1 500,00 €\nMais trabalho 1 25,59 € 25,59 €\nSubtotal: 1 525,59 €\nImposto (0%): 0,00 €\nTotal: 1 525,59 €",
   1 =>
     "✂\nReceipt Payment part Account / Payable to\nCH4431999123000889012\n✂\nMax Muster & Söhne\nAccount / Payable to\nCH4431999123000889012 Musterstrasse 123\nMax Muster & Söhne 8000 Seldwyla\nMusterstrasse 123\n8000 Seldwyla\nReference\n210000000003139471430009017\nReference\n210000000003139471430009017\nAdditional information\nBestellung vom 15.10.2020\nPayable by (name/address)\nSimon Muster\nPayable by (name/address)\nMusterstrasse 1\nCurrency Amount\nSimon Muster\n8000 Seldwyla\nCHF 1 949.75 Musterstrasse 1\n8000 Seldwyla\nCurrency Amount\nCHF 1 949.75\nAcceptance point"
 }}

Extract text from only some pages.

iex> content = File.read!("priv/fixtures/fatura.pdf")
...> PdfExtractor.extract_text_from_binary(content, [0])
{:ok,
 %{
   0 =>
     "Text Example Bill FATURA\n# 2025010002\nData: Jun 21, 2025\nProjeto de lei para:\nSaldo devedor: 1 525,59 €\nElixir Company\nItem Quantidade Avaliar Quantia\nTrabalho 1 1 500,00 € 1 500,00 €\nMais trabalho 1 25,59 € 25,59 €\nSubtotal: 1 525,59 €\nImposto (0%): 0,00 €\nTotal: 1 525,59 €"
 }}

Extract only the titles in the book.

iex> content = File.read!("priv/fixtures/frankenstein-mary-shelley.pdf")
...>
...> PdfExtractor.extract_text_from_binary(content, %{
...>   1 => {0, 0, 612, 190},
...>   2 => {0, 0, 612, 190},
...>   8 => {0, 0, 612, 190},
...>   12 => {0, 0, 612, 190},
...>   13 => {0, 0, 612, 190}
...> })
{:ok,
 %{
   1 => "Frankenstein",
   2 => "Introduction",
   8 => "Preface",
   12 => "Frankenstein",
   13 => "Letter I"
 }}

Extract multiple areas from a single page.

iex> content = File.read!("priv/fixtures/frankenstein-mary-shelley.pdf")
...>
...> PdfExtractor.extract_text_from_binary(content, %{
...>   13 => [{0, 200, 600, 300}, {0, 400, 612, 440}]
...> })
{:ok,
 %{
   13 => [
     "To Mrs. Saville, England.\nSt. Petersburgh, Dec. 11th, 17 —.",
     "I am already far north of London; and as I walk in the streets of\nPetersburgh, I feel a cold northern breeze play upon my cheeks, which"
   ]
 }}

Extracts words from PDF pages together with their position on the page.

Each word is a map with its :text and the edges of its bounding box, :left, :right, :top and :bottom, in PDF points measured from the top left corner of the page. Words come in the order pdfplumber reads them, which is the same order extract_text/3 uses to build its text.

See extract_text/3 for details on how to specify pages, areas, and options.

Options

Besides :timeout, :name and :password, this function accepts:

• :font - when true, adds the :font_name and :font_size of each word. The subset tag that PDFs prepend to the names of embedded fonts, such as ABCDEF+, is removed. A word written in more than one font is split where the font changes (default: false)

Examples

Extract the words of a single page.

iex> {:ok, %{0 => words}} = PdfExtractor.extract_words("priv/fixtures/fatura.pdf", 0)
...> Enum.take(words, 3)
[
  %{
    text: "Text",
    left: 48.0,
    right: 67.99875,
    top: 26.64224999999999,
    bottom: 36.39224999999999
  },
  %{
    text: "Example",
    left: 70.734375,
    right: 110.757,
    top: 26.64224999999999,
    bottom: 36.39224999999999
  },
  %{
    text: "Bill",
    left: 113.49609375,
    right: 128.63240625,
    top: 26.64224999999999,
    bottom: 36.39224999999999
  }
]

Extract only the words inside an area.

iex> PdfExtractor.extract_words("priv/fixtures/frankenstein-mary-shelley.pdf", %{
...>   1 => {0, 0, 612, 190}
...> })
{:ok,
 %{
   1 => [
     %{
       text: "Frankenstein",
       left: 76.99218429199217,
       right: 177.95707306012164,
       top: 102.2249957406251,
       bottom: 132.22499449062502
     }
   ]
 }}

Get the font of each word to tell headings from body text.

iex> PdfExtractor.extract_words(
...>   "priv/fixtures/frankenstein-mary-shelley.pdf",
...>   %{
...>     1 => {0, 0, 612, 190}
...>   },
...>   font: true
...> )
{:ok,
 %{
   1 => [
     %{
       text: "Frankenstein",
       left: 76.99218429199217,
       right: 177.95707306012164,
       top: 102.2249957406251,
       bottom: 132.22499449062502,
       font_name: "LeagueGothic-Regular",
       font_size: 29.999998749999918
     }
   ]
 }}

Extracts words from PDF binary data. See extract_words/3 for the shape of each word and extract_text/3 for details on how to specify pages, areas, and options.

Examples

Extract only the words inside an area.

iex> content = File.read!("priv/fixtures/frankenstein-mary-shelley.pdf")
...> PdfExtractor.extract_words_from_binary(content, %{1 => {0, 0, 612, 190}})
{:ok,
 %{
   1 => [
     %{
       text: "Frankenstein",
       left: 76.99218429199217,
       right: 177.95707306012164,
       top: 102.2249957406251,
       bottom: 132.22499449062502
     }
   ]
 }}

Renders PDF pages to PNG images.

Each page maps to the PNG binary of the page, or of the area when one is given, so the same function makes thumbnails of whole pages and crops of a single table or paragraph. Images are quantized to 256 colors, which keeps them small.

See extract_text/3 for details on how to specify pages and areas.

Options

Besides :timeout, :name and :password, this function accepts:

• :width - the width of the image in pixels. The height follows the proportions of the page
• :resolution - pixels per inch (default: 72, which gives one pixel per point)
• :antialias - when true, smooths the edges of text and lines (default: false)

:width and :resolution cannot be given together.

Examples

Render a thumbnail of the first page.

iex> {:ok, %{0 => png}} = PdfExtractor.render_pages("priv/fixtures/fatura.pdf", 0, width: 200)
...> <<_::binary-size(16), width::32, height::32, _::binary>> = png
...> {width, height}
{200, 259}

Render only the area of a table.

iex> {:ok, %{0 => png}} =
...>   PdfExtractor.render_pages("priv/fixtures/fatura.pdf", %{0 => {0, 180, 595, 260}},
...>     width: 200
...>   )
...>
...> <<_::binary-size(16), width::32, height::32, _::binary>> = png
...> {width, height}
{200, 26}

Renders pages from PDF binary data to PNG images. See render_pages/3 for its options and extract_text/3 for details on how to specify pages and areas.

Examples

iex> content = File.read!("priv/fixtures/fatura.pdf")
...> {:ok, %{0 => png}} = PdfExtractor.render_pages_from_binary(content, 0, width: 200)
...> <<_::binary-size(16), width::32, height::32, _::binary>> = png
...> {width, height}
{200, 259}