This article is part of the article series "MIT Introduction to Algorithms."
<- previous article next article ->

MIT AlgorithmsThis is the sixth post in an article series about MIT's lecture course "Introduction to Algorithms." In this post I will review lectures nine and ten, which are on the topic of Search Trees.

Search tree data structures provide many dynamic-set operations such as search, insert, delete, minimum element, maximum element and others. The complexity of these operations is proportional to the height of the tree. The better we can balance the tree, the faster they will be. Lectures nine and ten discuss two such structures.

Lecture nine discusses randomly built binary search trees. A randomly built binary search tree is a binary tree that arises from inserting the keys in random order into an initially empty tree. The key result shown in this lecture is that the height of this tree is O(lg(n)).

Lecture ten discusses red-black trees. A red-black tree is a binary search tree with extra bit of information at each node -- it's color, which can be either red or black. By contrasting the way nodes are colored on any path from the root to a leaf, red-black trees ensure that the tree is balanced, giving us guarantees that the operations on this tree will run on O(lg(n)) time!

PS. Sorry for being silent for the past two weeks. I am preparing for job interviews at a company starting with 'G' and it is taking all my time. ;)

Lecture 9: Randomly Built Binary Search Trees

Lecture nine starts with an example of good and bad binary search tree. Given a binary tree with n nodes, a good trees has height log(n) but the bad one has height close to n. As the basic operations on trees run in time proportional to the height of the tree, it's recommended that we build the good trees and not the bad ones.

Before discussing randomly built binary search trees, professor Erik Demaine shows another sorting algorithm. It's called binary search tree sort (BST-sort). It's amazingly simple -- given an array of n items to sort, build a BST out of it and do an in-order tree walk on it. In-order tree walk walks the left branch first, then prints the values, and then walks the right branch. Can you see why the printed list of values is sorted? (If not see the lecture ;) ) [part three of the article series covers sorting algorithms]

Turns out that there is a relation between BST-sort and quicksort algorithm. BST-sort and quicksort make the same comparisons but in different order. [more info on quicksort in part two of article series and in "three beautiful quicksorts" post]

After this discussion, the lecture finally continues with randomized BST-sort which leads to idea of randomly built BSTs.

The other half of the lecture is devoted to a complicated proof of the expected height of a randomly built binary search tree. The result of this proof is that the expected height is order log(n).

You're welcome to watch lecture nine:

Topics covered in lecture nine:

  • [00:50] Good and bad binary search trees (BSTs).
  • [02:00] Binary search tree sort tree algorithm.
  • [03:45] Example of running BST-sort on array (3, 1, 8, 2, 6, 7, 5).
  • [05:45] Running time analysis of BST-sort algorithm.
  • [11:45] BST-sort relation to quicksort algorithm.
  • [16:05] Randomized BST-sort.
  • [19:00] Randomly built binary search trees.
  • [24:58] Theorem: expected height of a rand BST tree is O(lg(n)).
  • [26:45] Proof outline.
  • [32:45] Definition of convex function.
  • [46:55] Jensen's inequality.
  • [55:55] Expected random BST height analysis.

Lecture nine notes:

MIT Algorithms Lecture 9 Notes Thumbnail. Page 1 of 2.
Lecture 9, page 1 of 2.

MIT Algorithms Lecture 9 Notes Thumbnail. Page 2 of 2.
Lecture 9, page 2 of 2.

Lecture 10: Red-Black Trees

Lecture ten begins with a discussion of balanced search trees. Balanced search tree is search tree data structure maintain a dynamic set of n elements using tree of height log(n).

There are many balanced search tree data structures. For example: AVL trees (invented in 1964), 2-3 trees (invented in 1970), 2-3-4 trees, B-trees, red-black trees, skiplists, treaps.

This lecture focuses exclusively on red-black trees.

Red-black trees are binary search trees with extra color field for each node. They satisfy red-black properties:

  • Every node is either red or black.
  • The root and leaves are black.
  • Every red node has a black parent.
  • All simple paths from a node to x to a descendant leaf of x have same number of black nodes = black-height(x).

The lecture gives a proof sketch of the height of an RB-tree and discusses running time of queries (search, min, max, successor, predecessor operations) and then goes into details of update operations (insert, delete). Along the way rotations on a tree are defined, the right-rotate and left-rotate ops.

The other half of the lecture looks at Red-Black-Insert operation that inserts an element in the tree while maintaining the red-black properties.

Here is the video of lecture ten:

Topics covered in lecture ten:

  • [00:35] Balanced search trees.
  • [02:30] Examples of balanced search tree data structures.
  • [05:16] Red-black trees.
  • [06:11] Red-black properties.
  • [11:26] Example of red-black tree.
  • [17:30] Height of red-black tree.
  • [18:50] Proof sketch of RBtree height.
  • [21:30] Connection of red-black trees to 2-3-4 trees.
  • [32:10] Running time of query operations.
  • [35:37] How to do RB-tree updates (inserts, deletes)?
  • [36:30] Tree rotations.
  • [40:55] Idea of red-black tree insert operation.
  • [44:30] Example of inserting an element in a tree.
  • [54:30] RB-Insert algorithm.
  • [01:03:35] The three cases in insert operation.

Lecture ten notes:

MIT Algorithms Lecture 10 Notes Thumbnail. Page 1 of 2.
Lecture 10, page 1 of 2.

MIT Algorithms Lecture 10 Notes Thumbnail. Page 2 of 2.
Lecture 10, page 2 of 2.

Have fun building trees! The next post will be about general methodology of augmenting data structures and it will discuss dynamic order statistics and interval trees!

PS. This course is taught from the CLRS book (also called "Introduction to Algorithms"):

Google ChromeAs everyone already knows, Google released a new open-source web browser called Chrome.

Having interest in code reuse, I downloaded the source code and examined all the open-source libraries used.

Google Chrome browser shows excellent example of code reuse. I found that they use at least 25 different software libraries!

Here is the full list of libraries, along with relative paths to source code and short library descriptions. Many of the libraries have been patched by googlers; look for files in each library directory for information about changes.


Relative Path


Google Breakpad


An open-source multi-platform crash reporting system.

Google URL


A small library for parsing and canonicalizing URLs.



Vector graphics engine.

Google v8


Google's open source JavaScript engine. V8 implements ECMAScript as specified in ECMA-262, 3rd edition, and runs on Windows XP and Vista, Mac OS X 10.5 (Leopard), and Linux systems that use IA-32 or ARM processors. V8 can run standalone, or can be embedded into any C++ application.



Open source web browser engine.

Netscape Portable Runtime (NSPR)


Netscape Portable Runtime (NSPR) provides a platform-neutral API for system level and libc like functions.

Network Security Services (NSS)


Network Security Services (NSS) is a set of libraries designed to support cross-platform development of security-enabled client and server applications. Applications built with NSS can support SSL v2 and v3, TLS, PKCS #5, PKCS #7, PKCS #11, PKCS #12, S/MIME, X.509 v3 certificates, and other security standards.



Spell checker and morphological analyzer library and program designed for languages with rich morphology and complex word compounding or character encoding.

Windows Template Library


C++ library for developing Windows applications and UI components. It extends ATL (Active Template Library) and provides a set of classes for controls, dialogs, frame windows, GDI objects, and more.

Google C++ Testing Framework


Google's framework for writing C++ tests on a variety of platforms (Linux, Mac OS X, Windows, Windows CE, and Symbian). Based on the xUnit architecture. Supports automatic test discovery, a rich set of assertions, user-defined assertions, death tests, fatal and non-fatal failures, various options for running the tests, and XML test report generation.

bsdiff and bspatch

/src/third_party/bsdiff and /src/third_party/bspatch

bsdiff and bspatch are tools for building and applying patches to binary files.



bzip2 compresses files using the Burrows-Wheeler block sorting text compression algorithm, and Huffman coding.

International Components for Unicode (ICU)


ICU is a mature, widely used set of C/C++ and Java libraries providing Unicode and Globalization support for software applications.



Library for handling the JPEG (JFIF) image format.



PNG image format library. It supports almost all PNG features, is extensible, and has been extensively tested for over 13 years.



XML C parsing library.



XSLT C library.



LZMA is the default and general compression method of 7z format in the 7-Zip program.



A collection of high performance c-string transformations (in this case, base 64 encoding/decoding), frequently 2x faster than standard implementations (if they exist at all).

Netscape Plugin Application Programming Interface (NPAPI)


Cross-platform plugin architecture used by many web browsers.



Application programming interface (API) for writing multithreaded applications

SCons - a software construction tool


Open Source software construction tool—that is, a next-generation build tool. Think of SCons as an improved, cross-platform substitute for the classic Make utility with integrated functionality similar to autoconf/automake and compiler caches such as ccache.



Software library that implements a self-contained, serverless, zero-configuration, transactional SQL database engine.

TLS Lite


Free Python library that implements SSL 3.0, TLS 1.0, and TLS 1.1. TLS Lite supports non-traditional authentication methods such as SRP, shared keys, and cryptoIDs in addition to X.509 certificates. Note: Python is not a part of Chrome. It's used for testing various parts of Chrome browser, such as code coverage, dependencies, measures page load times, compares generated html, etc.



zlib is designed to be a free, general-purpose, legally unencumbered -- that is, not covered by any patents -- lossless data-compression library for use on virtually any computer hardware and operating system.

They have done a really good job making these libraries work together. As someone said, "good coders code, great reuse."

I also found some other exciting things in the source, which I will soon post about. I recommend that you subscribe to my rss feed, if you are interested!

Talking about Chrome, I am waiting for Google to add capability to write extensions for their browser! I already made a list of extensions that I will try to create as soon as they add this feature.

This article is part of the article series "MIT Introduction to Algorithms."
<- previous article next article ->

MIT AlgorithmsThis is the fifth post in an article series about MIT's lecture course "Introduction to Algorithms." In this post I will review lectures seven and eight, which are on the topic of Hashing.

Many applications require a dynamic set that supports dictionary operations insert, search, and delete. For example, a compiler for a computer language maintains a symbol table, in which the keys of elements are arbitrary strings that correspond to identifiers in the language. A hash table is an effective data structure for implementing dictionaries.

Lectures seven and eight cover various implementation techniques of hash tables and hash functions.

Lecture 7: Hashing I

Lecture seven starts with the symbol-table problem -- given a table S, we'd like to insert an element into S, delete an element from S and search for an element in S. We'd also like these operations to take constant time.

The simplest solution to this problem is to use a direct-access (or direct-address) table. To represent the dynamic set, we use an array, or direct-address table, denoted by T, in which each position, or slot, corresponds to a key.

Using direct-address table, the dictionary operations are trivial to implement.

Direct-Address-Search(T, k)
    return T[k]

Direct-Address-Insert(T, x)
    T[key[x]] = x

Direct-Address-Delete(T, x)
    T[key[x]] = NIL

Direct addressing is applicable when we can afford to allocate an array that has one position for every possible key. It is not applicable when the range of keys can be large (as it requires a lot of space for the array T). This is where hashing comes in.

The lecture continues with explanation of what hashing is. Hashing uses a hash function h(k) that maps keys k randomly into slots of hash-table T. There is one hitch: two keys may hash to the same slot. We call this situation a collision. Fortunately, there are effective techniques for resolving the conflict created by collisions.

One of the simplest collision resolution techniques is called chaining. In chaining, we put all the elements that hash to the same slot in a linked list:


Collision resolution by chaining. Each hash-table slot T[j] contains a linked list of all the keys whose hash value is j. For example, h(k1) = h(k4).

Professor Leiserson then analyzes the running time of insert, delete and search operations. It is concluded that the expected running time operations is still O(1), under assumption that the number of hash-table slots is at least proportional to the number of elements in the table.

The other half of the lecture is devoted to hash functions and another way of resolving collisions -- resolving collisions by open addressing, and probing strategies (search) for open addressing -- linear probing and double hashing.

A good hash function should distribute the keys uniformly into the slots of the table and the regularity in the key distributions should not affect uniformity.

Two hash function creating methods are introduced - the division method, which defines h(k) = k mod m, and the multiplication method, where h(k) = (A·k mod 2w)>>(w-r), where w is bits in a word, A is an odd integer between 2w-1 and 2w, and r is lg(m).

You're welcome to watch lecture seven:

Topics covered in lecture seven:

  • [00:30] Symbol-table problem.
  • [02:05] Symbol-table operations: insert, delete, search.
  • [04:35] Direct-address table (direct addressing).
  • [09:45] Hashing.
  • [14:30] Resolving hash function collisions by chaining.
  • [17:05] Worst-case analysis of chaining.
  • [19:15] Average-case analysis of chaning.
  • [29:30] Choosing a hash function.
  • [30:55] Division method hash function.
  • [39:05] Multiplication method hash function.
  • [46:30] Multiplication method explained with a modular wheel.
  • [50:12] Resolving hash function collisions by open addressing.
  • [59:00] Linear probing strategy.
  • [01:01:30] Double hashing probing strategy.
  • [01:04:20] Average-case analysis of open addressing.

Lecture seven notes:

MIT Algorithms Lecture 7 Notes Thumbnail. Page 1 of 2.
Lecture 7, page 1 of 2.

MIT Algorithms Lecture 7 Notes Thumbnail. Page 2 of 2.
Lecture 7, page 2 of 2.

Lecture 8: Hashing II

Lecture eight starts with addressing a weakness of hashing -- for any choice of hash function, there exists a set of keys that all hash to the same value. This weakness can lead to denial of service attacks on the application using hashing.

The idea of addressing this problem is to choose a hash function at random! This is called universal hashing.

The lecture then moves to a mathematically rigorous the definition of universal hashing and explains one of many ways to construct a universal hash function.

The other half of the lecture is devoted to perfect hashing. Perfect hashing solves a problem of constructing a static hash table (such as a hash table stored on a CD), so that searches take O(1) time guaranteed (worst-case). The key idea in creating such hash table is to use 2-level universal hashing, so that no collisions occur in level 2.

Video of lecture eight:

Topics covered in lecture eight:

  • [00:30] Fundamental weakness of hashing.
  • [05:12] Universal hashing.
  • [20:10] Constructing a universal hash function.
  • [49:45] Perfect hashing.
  • [54:30] Example of perfect hashing.
  • [01:06:27] (Markov inequality)
  • [01:14:30] Analysis of storage requirements for perfect hashing.

Lecture eight notes:

MIT Algorithms Lecture 8 Notes Thumbnail. Page 1 of 2.
Lecture 8, page 1 of 2.

MIT Algorithms Lecture 8 Notes Thumbnail. Page 2 of 2.
Lecture 8, page 2 of 2.

Have fun hashing! The next post will be about random binary search trees and red-black trees!

PS. This course is taught from the CLRS book (also called "Introduction to Algorithms"):

Linux ifconfig GolfingA friend of mine recently showed me his version of extracting ip address(es) from ifconfig. The ifconfig tool on Linux is used to list IP addresses and configure network interfaces. It is sometimes necessary to capture your IP in a variable, like when writing some firewall rules.

I decided to show off and golf the extraction of IP address with four commonly used tools -- Awk, sed, a version of Perl everyone has, and the latest version of Perl 5.10.

His original version was pretty ugly:

$ ifconfig | perl -ple 'print $_ if /inet addr/ and $_ =~ s/.*inet addr:((?:\d+\.){3}\d+).*/$1/g  ;$_=""' | grep -v ^\s*$

My own version on my firewall was:

$ ifconfig | grep inet | awk -F: '{ print $2 }' | awk '{ print $1 }'

This looks nicer but I was uselessly using grep, and calling awk twice.

Golfing in Perl

My first attempt was to do it with grep and Perl.

$ ifconfig | grep inet | perl -ple '($_) = /addr:([^ ]+)/'

In a split second I realized that Perl does grep itself:

$ ifconfig | perl -nle '/addr:([^ ]+)/ and print $1'

Then I noticed that 'dr:' matches the same lines as 'addr:'; also 'and' can be replaced with '&&':

$ ifconfig | perl -nle '/dr:([^ ]+)/ && print $1'

The regular expression '([^ ]+)' can be replaced with '(\S+)', which matches non-whitespace characters:

$ ifconfig | perl -nle '/dr:(\S+)/ && print $1'

Cutting the whitespace, the final version for Perl is 37 characters long:

$ ifconfig|perl -nle'/dr:(\S+)/&&print$1'

Golfing in Perl 5.10

It can be made shorter with Perl 5.10, which introduces the new say function:

$ ifconfig|perl -nE'/dr:(\S+)/&&say$1'

The result for Perl 5.10 is 34 characters.

Golfing in sed

Then I also tried doing the same with sed:

$ ifconfig | sed -n '/dr:/{;s/.*dr://;s/ .*//;p;}'

I modified it to strip off everything that was not numbers and dots:

$ ifconfig | sed -n '/dr:/{;s/.*dr:\([0-9.]\+\) .*/\1/;p;}'

This turned out to be much longer, so I tried getting rid of backslashes, by enabling extended regexes in sed with -r argument:

$ ifconfig | sed -rn '/dr:/{;s/.*dr:([0-9.]+) .*/\1/;p;}'

I forgot that I had used the ([^ ]+) regex before. Another my friend reminded me this, shortening the sed script to:

$ ifconfig | sed -rn 's/.*r:([^ ]+) .*/\1/p'

Dropping the whitespace, sed version turned out to be 40 characters long.

Golfing in Awk

My final attempt was to optimize my original Awk version:

$ ifconfig | awk '/dr:/{gsub(/.*:/,"",$2);print$2}'

Cutting the whitespace, the Awk version is 43 characters.

The Winner

The winner in this golf tournament is Perl 5.10 with 34 chars!

Can You Do Better?

Can you golf this even shorter?

PS. my awk and sed cheat sheets might come handy!

GNU Awk YouTube Downloader RevisitedAround a year ago I wrote a YouTube video downloader in GNU Awk. As I explained, I did it to explore the corners of Awk language.

The key idea in writing this program was to figure out how to do networking in Awk. The original version of Awk does not have any networking capabilities, but it turned out that GNU version of Awk does! There is even a manual on "TCP/IP Networking With Gnu Awk"!

One of my blog readers, Werner Illchmann, suggested that I add a progress bar and sent me a patch. I improved it a little and here is the result:

$ chmod +x get_youtube_vids.awk
$ ./get_youtube_vids.awk
Parsing YouTube video urls/IDs...
Getting video information for video: 4bQOSRm9YiQ...
Downloading: Premature Optimization is the Root of All Evil!...
Saving video to file 'Premature_Optimization_is_the_Root_of_All_Evil.flv' (size: 227.81kb)...
<strong>Done: 85121/233280 bytes (36%, 83.13kb/227.81kb)</strong>

Here is the source code of the program:

#!/usr/bin/gawk -f
# 2007.07.10 v1.0 - initial release
# 2007.10.21 v1.1 - youtube changed the way it displays vids
# 2008.03.01 v1.2 - youtube changed the way it displays vids
# 2008.08.28 v1.3 - added a progress bar and removed need for --re-interval 
# 2009.08.25 v1.4 - youtube changed the way it displays vids
# Peteris Krumins (
# -- good coders code, great reuse
# Usage: gawk -f get_youtube_vids.awk < | ID1> ...
# or just ./get_youtube_vids.awk < | ID1>

    if (ARGC == 1) usage();

    BINMODE = 3

    delete ARGV[0]
    print "Parsing YouTube video urls/IDs..."
    for (i in ARGV) {
        vid_id = parse_url(ARGV[i])
        if (length(vid_id) < 6) { # havent seen youtube vids with IDs < 6 chars
            print "Invalid YouTube video specified: " ARGV[i] ", not downloading!"
        VIDS[i] = vid_id

    for (i in VIDS) {
        print "Getting video information for video: " VIDS[i] "..."
        get_vid_info(VIDS[i], INFO)

        if (INFO["_redirected"]) {
            print "Could not get video info for video: " VIDS[i]

        if (!INFO["video_url"]) {
            print "Could not get video_url for video: " VIDS[i]
            print "Please goto my website, and submit a comment with an URL to this video, so that I can fix it!"
            print "Url:"
        if ("title" in INFO) {
            print "Downloading: " INFO["title"] "..."
            title = INFO["title"]
        else {
            print "Could not get title for video: " VIDS[i]
            print "Trying to download " VIDS[i] " anyway"
            title = VIDS[i]
        download_video(INFO["video_url"], title)

function usage() {
    print "Downloading YouTube Videos with GNU Awk"
    print "Peteris Krumins ("
    print "  --  good coders code, great reuse"
    print "Usage: gawk -f get_youtube_vids.awk < | ID1> ..."
    print "or just ./get_youtube_vids.awk < | ID1> ..."
    exit 1

# function parse_url
# takes a url or an ID of a youtube video and returns just the ID
# for example the url could be the full url:
# or it could be
# or just or
# or just the ID
function parse_url(url) {
    gsub(/http:\/\//, "", url)                # get rid of http:// part
    gsub(/www\./,     "", url)                # get rid of www.    part
    gsub(/youtube\.com\/watch\?v=/, "", url)  # get rid of part

    if ((p = index(url, "&")) > 0)      # get rid of &foo=bar&... after the ID
        url = substr(url, 1, p-1)

    return url

# function get_vid_info
# function takes the youtube video ID and gets the title of the video
# and the url to .flv file
function get_vid_info(vid_id, INFO,    InetFile, Request, HEADERS, matches, escaped_urls, fmt_urls, fmt) {
    delete INFO
    InetFile = "/inet/tcp/0/"
    Request = "GET /watch?v=" vid_id " HTTP/1.1\r\n"
    Request = Request "Host:\r\n\r\n"

    get_headers(InetFile, Request, HEADERS)
    if ("Location" in HEADERS) {
        INFO["_redirected"] = 1

    # fix this bug:
    while ((InetFile |& getline) > 0) {
        if (match($0, /"fmt_url_map": "([^"]+)"/, matches)) {
            escaped_urls = url_unescape(matches[1])
            split(escaped_urls, fmt_urls, /,?[0-9]+\|/)
            for (fmt in fmt_urls) {
                if (fmt_urls[fmt] ~ /itag=5/) {
                    # fmt number 5 is the best video
                    INFO["video_url"] = fmt_urls[fmt]
        else if (match($0, /<title>YouTube - ([^<]+)</, matches)) {
            # lets try to get the title of the video from html tag which is
            # less likely a subject to future html design changes
            INFO["title"] = matches[1]

# function url_unescape
# given a string, it url-unescapes it.
# charactes such as %20 get converted to their ascii counterparts.
function url_unescape(str,    nmatches, entity, entities, seen, i) {
    nmatches = find_all_matches(str, "%[0-9A-Fa-f][0-9A-Fa-f]", entities)
    for (i = 1; i <= nmatches; i++) {
        entity = entities[i]
        if (!seen[entity]) {
            if (entity == "%26") { # special case for gsub(s, r, t), when r = '&'
                gsub(entity, "\\&", str)
            else {
                gsub(entity, url_entity_unescape(entity), str)
            seen[entity] = 1
    return str

# function find_all_matches
function find_all_matches(str, re, arr,    j, a, b) {
    a = RSTART; b = RLENGTH   # to avoid unexpected side effects

    while (match(str, re) > 0) {
        arr[++j] = substr(str, RSTART, RLENGTH)
        str = substr(str, RSTART+RLENGTH)
    RSTART = a; RLENGTH = b
    return j

# function url_entity_unescape
# given an url-escaped entity, such as %20, return its ascii counterpart.
function url_entity_unescape(entity) {
    sub("%", "", entity)
    return sprintf("%c", strtonum("0x" entity))

# function download_video
# takes the url to video and saves the movie to current directory using
# santized video title as filename
function download_video(url, title,    filename, InetFile, Request, Loop, HEADERS, FOO) {
    title = sanitize_title(title)
    filename = create_filename(title)

    parse_location(url, FOO)
    InetFile = FOO["InetFile"]
    Request  = "GET " FOO["Request"] " HTTP/1.1\r\n"
    Request  = Request "Host: " FOO["Host"] "\r\n\r\n"

    Loop = 0 # make sure we do not get caught in Location: loop
    do {     # we can get more than one redirect, follow them all
        get_headers(InetFile, Request, HEADERS)
        if ("Location" in HEADERS) { # we got redirected, let's follow the link
            parse_location(HEADERS["Location"], FOO)
            InetFile = FOO["InetFile"]
            Request  = "GET " FOO["Request"] " HTTP/1.1\r\n"
            Request  = Request "Host: " FOO["Host"] "\r\n\r\n"
            if (InetFile == "") {
                print "Downloading '" title "' failed, couldn't parse Location header!"
    } while (("Location" in HEADERS) && Loop < 5)

    if (Loop == 5) {
        print "Downloading '" title "' failed, got caught in Location loop!"
    print "Saving video to file '" filename "' (size: " bytes_to_human(HEADERS["Content-Length"]) ")..."
    save_file(InetFile, filename, HEADERS)
    print "Successfully downloaded '" title "'!"

# function sanitize_title
# sanitizes the video title, by removing ()'s, replacing spaces with _, etc.
function sanitize_title(title) {
    gsub(/\(|\)/, "", title)
    gsub(/[^[:alnum:]-]/, "_", title)
    gsub(/_-/, "-", title)
    gsub(/-_/, "-", title)
    gsub(/_$/, "", title)
    gsub(/-$/, "", title)
    gsub(/_{2,}/, "_", title)
    gsub(/-{2,}/, "-", title)
    return title

# function create_filename
# given a sanitized video title, creates a nonexisting filename
function create_filename(title,    filename, i) {
    filename = title ".flv"
    i = 1
    while (file_exists(filename)) {
        filename = title "-" i ".flv"
    return filename

# function save_file
# given a special network file and filename reads from network until eof
# and saves the read contents into a file named filename
function save_file(Inet, filename, HEADERS,    done, cl, perc, hd, hcl) {
    OLD_RS  = RS

    ORS = ""

    # clear the file
    print "" > filename

    # here we will do a little hackery to write the downloaded data
    # to file chunk by chunk instead of downloading it all to memory
    # and then writing
    # the idea is to use a regex for the record field seperator
    # everything that gets matched is stored in RT variable
    # which gets written to disk after each match
    # RS = ".{1,512}" # let's read 512 byte records

    RS = "@" # I replaced the 512 block reading with something better.
             # To read blocks I had to force users to specify --re-interval,
             # which made them uncomfortable.
             # I did statistical analysis on YouTube video files and
             # I found that hex value 0x40 appears pretty often (200 bytes or so)!

    cl = HEADERS["Content-Length"]
    hcl = bytes_to_human(cl)
    done = 0
    while ((Inet |& getline) > 0) {
        done += length($0 RT)
        perc = done*100/cl
        hd = bytes_to_human(done)
        printf "Done: %d/%d bytes (%d%%, %s/%s)            \r",
            done, cl, perc, bytes_to_human(done), bytes_to_human(cl)
        print $0 RT >> filename
    printf "Done: %d/%d bytes (%d%%, %s/%s)            \n",
        done, cl, perc, bytes_to_human(done), bytes_to_human(cl)

    RS  = OLD_RS

# function get_headers
# given a special inet file and the request saves headers in HEADERS array
# special key "_status" can be used to find HTTP response code
# issuing another getline() on inet file would start returning the contents
function get_headers(Inet, Request,    HEADERS, matches, OLD_RS) {
    delete HEADERS

    # save global vars

    print Request |& Inet

    # get the http status response
    if (Inet |& getline > 0) {
        HEADERS["_status"] = $2
    else {
        print "Failed reading from the net. Quitting!"
        exit 1

    while ((Inet |& getline) > 0) {
        # we could have used FS=": " to split, but i could not think of a good
        # way to handle header values which contain multiple ": "
        # so i better go with a match
        if (match($0, /([^:]+): (.+)/, matches)) {
            HEADERS[matches[1]] = matches[2]
        else { break }

# function parse_location
# given a Location HTTP header value the function constructs a special
# inet file and the request storing them in FOO
function parse_location(location, FOO) {
    # location might look like
    if (match(location, /http:\/\/([^\/]+)(\/.+)/, matches)) {
        FOO["InetFile"] = "/inet/tcp/0/" matches[1] "/80"
        FOO["Host"]     = matches[1]
        FOO["Request"]  = matches[2]
    else {
        FOO["InetFile"] = ""
        FOO["Host"]     = ""
        FOO["Request"]  = ""

# function bytes_to_human
# given bytes, converts them to human readable format like 13.2mb
function bytes_to_human(bytes,    MAP, map_idx, bytes_copy) {
    MAP[0] = "b"
    MAP[1] = "kb"
    MAP[2] = "mb"
    MAP[3] = "gb"
    MAP[4] = "tb"
    map_idx = 0
    bytes_copy = int(bytes)
    while (bytes_copy > 1024) {
        bytes_copy /= 1024

    if (map_idx > 4)
        return sprintf("%d bytes", bytes, MAP[map_idx])
        return sprintf("%.02f%s", bytes_copy, MAP[map_idx])

# function file_exists
# given a path to file, returns 1 if the file exists, or 0 if it doesn't
function file_exists(file,    foo) {
    if ((getline foo <file) >= 0) {
        return 1
    return 0

If you decide to learn Awk programming language, I suggest that you take a look at the Awk Cheat Sheet that I have made.

Download GNU Awk YouTube Video Downloader

Download link: gawk youtube video downloader
Total downloads: 27184 times