Research PDF Reader

Search a PDF for as many terms as you like at once, each in its own colour, and every spelling of each. Highlight and comment as you read. Everything happens on this computer: nothing is uploaded, and your notes stay in this browser.

Your book needs a text layer. If it has none, run it through gImageReader first, which is free, open source and entirely local; Book Scan Cleaner tidies the scan before that.

Drop a PDF here

or

How this works

Drop in a PDF that already carries a text layer. It is read once, on this computer, and nothing is uploaded. Close the tab and the book is gone; only your own highlights and comments are kept, in this browser.

Terms

Each row is one term with its own colour, and there can be as many rows as you like. They need not be related to each other: a family name in yellow beside a people or a place in red is a normal way to work. The number beside a row is how often it occurs in the whole book. Untick show to put a row aside without losing it.

How a term is matched

Case and diacritics never count: Modić, modic and MODIĆ are one and the same. What you choose per row is how much of a word has to match.

starts with
The default, and the one that handles Latin declension. Modich also finds Modicha and Modichem, but not quo modi.
whole word
Only the word itself, nothing grown onto it.
anywhere
Also inside another word, so Modich finds deModich. This is the one that drags in noise.

The choice matters more than it looks. modi is the genitive of modus and stands on nearly every page of a charter collection, so searching for it anywhere returns hundreds of hits of which a handful are the family.

Spellings

Under a row are variants the book itself contains, with their counts. They are found by the orthography of Latin charters — w, u and v as one letter, y for i, ch, cz, cs and ty for c, th for t, a doubled consonant for a single one — and beyond that only a letter that stands for another may differ: a vowel for a vowel, which is the scribe, or a pair a reading engine mixes up, which is the OCR. A difference at the very first letter is real but rarer, so it is offered last. An ending never counts, because that is what starts with is already for. So Mudich and Nodich are offered for Modich, and woman is not offered for Roman. Nothing is invented either: a form no page carries is never offered. Tick the ones that belong to your term and they are counted together, under one colour. What the list does not offer you can type yourself with + spelling: Roma and Rome under the same colour is a normal way to work. Click a spelling to drop it again.

Reading

Every occurrence is highlighted at once, and a name broken across a line is highlighted whole, on both lines. ‹ › beside the count walks through the hits in reading order; Ctrl+Shift+Left and Ctrl+Shift+Right do the same from the keyboard. The strip down the right edge is the whole book at a glance, one tick per hit in its row's colour. Where two colours sit at the same height, both terms are on that page: click the strip to go there.

Every mention in one list

All mentions opens a panel down the right side holding every hit of every showing term, in reading order, each with its page number and half a line either side. The ≡ beside a term narrows the same panel to that one term, its added spellings included; All terms in the panel widens it again. Click a line and the page opens at it. It is the way to read what a name is doing across a book without scrolling for it.

Highlights and comments

Select any words on a page, from one word to a whole paragraph, even one that runs onto the next page. A small bar appears beside them: pick a colour to highlight them, or Comment to write a note on them. Click a highlight to change its colour, give it a label such as Important or To verify, write its comment, or delete it. A label stays with its colour in this book: give one orange highlight Important and every orange highlight after it is Important too, until you choose another. A highlight that carries a comment has a small dot at its end.

Notes in the top bar lists every highlight and comment in reading order. Show only one colour or one label; click a note and the page opens at it. Export as Markdown writes them out as quotations with their pages and your comments, ready to go into a paper or a notebook.

Notes are kept in this browser, filed under the PDF's own fingerprint, so the same file opened again brings them back. They are never written into the PDF, and clearing the browser's data clears them too. Save notes writes them to a small file to keep, to take to another computer or to send to a colleague; Load notes adds such a file to what is already here.

Sets

save this set writes the rows to a small .json file — terms, colours, match modes. Load it on the next book, or send it to a colleague.

What it will not do

It does not recognise text. If the pages are pictures with no text layer, it says so and stops; run the book through OCR first. It never writes to your PDF: your highlights and comments are kept beside the file, never in it.