C$50 Finance

tl;dr

Implement a website via which users can "buy" and "sell" stocks, a la the below.

Background

If you’re not quite sure what it means to buy and sell stocks (i.e., shares of a company), head to http://www.investopedia.com/university/stocks/ for a tutorial.

You’re about to implement C$50 Finance, a web app via which you can manage portfolios of stocks. Not only will this tool allow you to check real stocks' actual prices and portfolios' values, it will also let you buy (okay, "buy") and sell (okay, "sell") stocks by querying Yahoo Finance for stocks' prices.

Indeed, Yahoo lets you download stock quotes in CSV format via URLs like http://download.finance.yahoo.com/d/quotes.csv?s=NFLX&f=sl1d1t1c1ohgv&e=.csv. Notice how Netflix’s symbol (NFLX) is embedded in this URL (as the value of the HTTP parameter called s); that’s how Yahoo knows whose data to return. Notice also the value of the HTTP parameter called f; it’s a bit cryptic (and officially undocumented), but the value of that parameter tells Yahoo which fields of data to return to you. If curious as to what they mean, here’s a handy reference. Incidentally, Yahoo sometimes returns prices in currencies other than US dollars (USD) without telling you which, but we’ll assume everything’s in US dollars for simplicity.

Let’s turn our attention now to the app’s distribution code!

Distribution

Downloading

$ wget https://github.com/cs50/problems/archive/finance.zip
$ unzip finance.zip
$ rm finance.zip
$ mv problems-finance finance
$ cd finance
$ ls
application.py  helpers.py        static
finance.db      requirements.txt  templates

Running

  1. Install this problem’s dependencies (within finance/):

    pip3 install --user -r requirements.txt
  2. Start Flask’s built-in web server (within finance/):

    flask run

    Select CS50 IDE > Web Server to see the distribution code in action. You won’t be able to log in or register, though, just yet!

  3. In another terminal window, start phpLiteAdmin (within finance/):

    phpliteadmin finance.db

    And open the URL that phpliteadmin outputs in a new tab. Notice how finance.db comes with a table called users. Take a look at its structure (i.e., schema). Notice how, by default, new users will receive $10,000 in cash. But there aren’t (yet!) any users (i.e., rows) therein to browse.

    Here on out, if you’d prefer a command line, you’re welcome to use sqlite3 instead of phpLiteAdmin.

Understanding

application.py

Open up application.py. Atop the file are a bunch of imports, among them CS50’s SQL module and a few helper functions. More on those soon.

After configuring Flask, notice how this file disables caching of responses (provided you’re in debugging mode, which you are by default on CS50 IDE), lest you make a change to some file but your browser not notice. Notice next how it configures Jinja with a custom "filter," usd, a function (defined in helpers.py) that will make it easier to format values as US dollars (USD). It then further configures Flask to store sessions on the local filesystem (i.e., disk) as opposed to storing them inside of (digitally signed) cookies, which is Flask’s default. The file then configures CS50’s SQL module to use finance.db, a SQLite database whose contents we’ll soon see!

Thereafter are a whole bunch of routes, only two of which are fully implemented: login and logout. Read through the implementation of login first. Notice how it uses db.execute (from CS50’s library) to query finance.db. And notice how it uses pwd_context.verify to compare hashes of users' passwords. Finally, notice how login "remembers" that a user is logged in by storing his or her user_id, an INTEGER, in session. That way, any of this file’s routes can check which user, if any, is logged in. Meanwhile, notice how logout simply clears session, effectively logging a user out.

Notice how most routes are "decorated" with @login_required (a function defined in helpers.py too). That decorator ensures that, if a user tries to visit any of those routes, he or she will first be redirected to login so as to log in.

Notice too how most routes support GET and POST. Even so, most of them (for now!) simply return an "apology," since they’re not yet implemented.

helpers.py

Next take a look at helpers.py. Ah, there’s the implementation of apology. Notice how it ultimately renders a template, apology.html. It also happens to define within itself another function, escape, that it simply uses to replace special characters in apologies. By defining escape inside of apology, we’ve scoped the former to the latter alone; no other functions will be able (or need) to call it.

Next in the file is login_required. No worries if this one’s a bit cryptic, but if you’ve ever wondered how a function can return another function, here’s an example!

Thereafter is lookup, a function that, given a symbol (e.g., NFLX), returns a stock quote for that symbol in the form of a dict with three keys: name, whose value is a str (e.g., "Netflix, Inc."); price, whose value is a float; and symbol, whose value is a str, a canonicalized (uppercase) version of a stock’s symbol, irrespective of how that symbol was capitalized when passed into lookup.

Last in the file is usd, a short function that simply formats a float as USD (e.g., 1234.56 is formatted as $1,234.56).

requirements.txt

Next take a quick look at requirements.txt. That file simply prescribes the packages on which this app will depend.

static/

Glance too at static/, inside of which is styles.css. That’s where some initial CSS lives. You’re welcome to alter it as you see fit.

templates/

Now look in templates/. In login.html is, essentially, just an HTML form, stylized with Bootstrap. In apology.html, meanwhile, is a template for an apology. Recall that apology in helpers.py took two arguments: top and bottom. Here they are again! And here’s why. 0:-)

Last up is layout.html. It’s a bit bigger than usual, but that’s mostly because it comes with a fancy, mobile-friendly "navbar" (navigation bar), also based on Bootstrap. Notice how it defines a block, main, inside of which templates (including apology.html and login.html) shall go. It also includes support for Flask’s message flashing so that you can relay messages from one route to another for the user to see.

Specification

register

Complete the implementation of register in such a way that it allows a user to register for an account.

  • Require that a user input a username. Render an apology if the user’s input is blank or the username already exists.

  • Require that a user input a password and then that same password again. Render an apology if either input is blank or the passwords do not match.

  • INSERT the new user into users, storing a hash of the user’s password, not the password itself. Odds are you’ll find pwd_context.hash of interest.

  • Odds are you’ll want to create a new template (e.g., register.html) that’s quite similar to login.html.

Once you’ve implemented register correctly, you should be able to register for an account and log in (since login and logout already work)! And you should be able to see your rows via phpLiteAdmin or sqlite3.

quote

Complete the implementation of quote in such a way that it allows a user to look up a stock’s current price.

  • Require that a user input a stock’s symbol.

  • Odds are you’ll want to create two new templates (e.g., quote.html and quoted.html). When a user visits /quote via GET, render one of those templates, inside of which should be an HTML form that submits to /quote via POST. In response to a POST, quote can render that second template, embedding within it one or more values from lookup.

buy

Complete the implementation of buy in such a way that it enables a user to buy stocks.

  • Require that a user input a stock’s symbol. Render an apology if the input is blank or the symbol does not exist (as per the return value of lookup).

  • Require that a user input a number of shares. Render an apology if the input is not a positive integer.

  • Odds are you’ll want to call lookup to look up a stock’s current price.

  • Odds are you’ll want to SELECT how much cash the user currently has in users.

  • Add one or more new tables to finance.db via which to keep track of the purchase. Store enough information so that you know who bought what at what price and when.

    • Use appropriate SQLite types.

    • Define UNIQUE indexes on any fields that should be unique.

    • Define (non-UNIQUE) indexes on any fields via which you will search (as via SELECT with WHERE).

  • Render an apology, without completing a purchase, if the user cannot afford the number of shares at the current price.

  • You don’t need to worry about race conditions (or use transactions).

Once you’ve implemented buy correctly, you should be able to see users' purchases in your new table(s) via phpLiteAdmin or sqlite3.

index

Complete the implementation of index in such a way that it displays an HTML table summarizing, for the user currently logged in, which stocks the user owns, the numbers of shares owned, the current price of each stock, and the total value of each holding (i.e., shares times price). Also display the user’s current cash balance along with a grand total (i.e., stocks' total value plus cash).

  • Odds are you’ll want to execute multiple SELECTs. Depending on how you implement your table(s), you might find GROUP BY, HAVING, SUM, and/or WHERE of interest.

  • Odds are you’ll want to call lookup for each stock.

sell

Complete the implementation of sell in such a way that it enables a user to sell shares of a stock (that he or she owns).

  • The user interface is entirely up to you. You may allow the user to sell one or more shares, or you may require the user to sell all shares of a stock at once.

  • You must check for and render apologies for any possible errors.

  • You don’t need to worry about race conditions (or use transactions).

history

Complete the implementation of history in such a way that it displays an HTML table summarizing all of a user’s transactions ever, listing row by row each and every buy and every sell.

  • For each row, make clear whether a stock was bought or sold and include the stock’s symbol, the (purchase or sale) price, the number of shares bought or sold, and the date and time at which the transaction occurred.

  • You might need to alter the table you created for buy or supplement it with an additional table. Try to minimize redundancies.

personal touch

Implement one (1) additional feature, a personal touch of your choice:

  • Empower users to change their passwords.

  • Empower users to add additional cash to their account.

  • Empower users to buy more shares or sell shares of stocks they already own via index itself, without having to type stocks' symbols manually.

  • Empower users with some other feature of comparable scope.

Walkthroughs

Testing

No check50 for this one, but be sure to try to "break" your own site, as by

  • inputting alpabetical strings into forms when only numbers are expected,

  • inputting zero or negative numbers into forms when only positive numbers are expected,

  • inputting floating-point values into forms when only integers are expected,

  • trying to spend more cash than a user has,

  • trying to sell more shares than a user has,

  • inputting an invalid stock symbol, and

  • including potentially dangerous characters like ' and ; in SQL queries.

Staff’s Solution

You’re welcome to stylize your own app differently, but here’s what the staff’s solution looks like!

Feel free to register for an account and play around. Do not use a password that you use on other sites.

It is reasonable to look at the staff’s HTML and CSS.

Hints

  • Within cs50.SQL is an execute method whose first argument should be a str of SQL. If that str contains named parameters to which values should be bound, those values can be provided as additional named parameters to execute. See the implementation of login for one such example. The return value of execute is as follows:

    • If str is a SELECT, then execute returns a list of zero or more dict objects, inside of which are keys and values representing a table’s fields and cells, respectively.

    • If str is an INSERT, and the table into which data was inserted contains an autoincrementing PRIMARY KEY, then execute returns the value of the newly inserted row’s primary key.

    • If str is a DELETE or an UPDATE, then execute returns the number of rows deleted or updated by str.

    If an INSERT or UPDATE would violate some constraint (e.g., a UNIQUE index), then execute returns None. In cases of error, execute raises a RuntimeError.

  • Be sure to use named bind parameters (i.e., a paramstyle of named) when calling CS50’s execute method, a la WHERE name=:name. Do not use format or + (i.e., concatenation), lest you risk a SQL injection attack.

  • If (and only if) already comfortable with SQL, you’re welcome to use SQLAlchemy Core or Flask-SQLAlchemy (i.e., SQLAlchemy ORM) instead of cs50.SQL.

  • You’re welcome to add additional static files to static/.

  • You’re welcome to select an alternative theme for your site as well as change the site’s aesthetics more generally, including apologies.

  • Odds are you’ll want to consult Jinja’s documentation when implementing your templates.

  • It is reasonable to ask others to try out (and try to trigger errors in) your site. Via Share in CS50 IDE’s top-right corner can you share your Application by making it Public. Take care not to share your Editor, which would provide access to your Python code and SQLite database.

FAQs

ImportError: No module named 'application'

By default, flask looks for a file called application.py in your current working directory (because we’ve configured the value of FLASK_APP, an environment variable, to be application.py). If seeing this error, odds are you’ve run flask in the wrong directory!

OSError: [Errno 98] Address already in use

If, upon running flask, you see this error, odds are you (still) have flask running in another tab. Be sure to kill that other process, as with ctrl-c, before starting flask again. If you haven’t any such other tab, execute fuser -k 8080/tcp to kill any processes that are (still) listening on TCP port 8080.

CHANGELOG

  • 2016-10-28

    • Initial release.