crossdelenna/tts_sample · Datasets at Hugging Face

text stringlengths 42 144	audio audioduration (s) 2.83 8
Open the web page you want to scrape in your browser. For this video I am going to scrape sgx
You know 8 or 9 years ago when I transitioned from discretionary to quantitative trading
response code in PyChimp's variable window here. A 200 response code means
the request was successful. indicates client error
indicates the request failed. Time to post process that FacedHTML data. Expanding the fetched data reveals a text
field containing an HTML formatted string. Beautiful soup then parses and cleans
this HTML using two lines of code. So pandas can process it with
df=pd.read_html(html_data) Pandas can convert this HTML
data into a neat data frame. Let's look at the special variable window again.
Here we can see that our data frame df actually has 9 different data frames within it.
If we expand it we can see that various data frames. Let's open the first data frame by clicking on View as data frame.
This is the same data frame that we can see in table format on the website. And if we click on the second data frame
I gradually realized that data is far more important than analytical skills.
it corresponds to the second table on the hgx nifty webpage.
If you want to fetch other indexes like the snp500 or laujones you can click on data frame
number 6 and get the relevant data. You can use the same code for various other symbols.
Since we need the hgx data we would use the first element in the df list.
Now if we open this hgx data frame a datetime index is missing. So we will create a column called dat.
Set it as the index and then export it as a CSV file. The remaining part of the code is optional.
Here I converted the hgx data frame into a human readable text format. Then used gtts and pygame
library to read aloud that text. Most examples for gtts use reading and writing files to the disk.
But here we wrote the files in virtual memory to speed up the process. And if you have noticed I am using a python environment
installed on raspberry pi via sssg in pycharm. If you want to know more about how to do that I have made a video on it.
I mean look at this way. There is an upper limit to how much analysis you can perform on a single data source.
Check that out. I will put the link in the description. If I execute these lines you will hear the tts engine.
I have connected the speaker to raspberry pi's auxiliary input. Sound quality might sound different
because my microphone is far away. Listen.
So we will conclude our video with this. Our job is done. We will put this script in a linux cron tab and if you are on
windows you can put the script in windows task scheduler. In the next video I will show you how I modified the investing.py
library using the same technique explained in this video. That library is down because of a cloudflare firewall.
We will use our adjusted script to grab the economic event data. After that we will feed the human readable version into
google's text to speech engine like we did here. So take care and happy coding. See you in the next one.
But imagine finding an underutilized data source and then applying similar advanced analytical methods and
seeking that crucial edge. Furthermore data has become crucial for machine learning or AI.
Therefore innovative ways to find data where others aren't looking are increasingly important.
We can't rely on traditional market data vendors for the same old generic data everyone else uses.
This is where web scraping becomes valuable. So let's get into a practical example. Our first web scraping
script will fetch SGX Nifty data. So this is the website we are going to scrape the data from.
The site uses other symbols so you can easily modify a small
nifty data as an example and I am using Edge a chromium based browser.
part of the code to use them instead. The basic structure of Python code for web scraping is
straightforward and largely consistent. The real trick lies in obtaining the
necessary URLs and headers so your script mimics like a web browser.
Please consider the following disclaimer before proceeding. One crucial thing to remember is the
legality of web scraping. It's generally acceptable to scrape
openly available data but don't redistribute commercially or collect personal information like
email IDs or user names. I mean there is a still grey area but avoid overloading the target servers.
This can lead to your IP being banned. Keeping these points
If you want to remember only one thing from this video then remember this. You simply have to right click on the
webpage you want to scrape the data from and select inspect element from the popup menu. That's all.
So the method is virtually identical to Google Chrome. Simply press Ctrl plus Shift plus I or
Alternatively use the keyboard shortcut Ctrl Shift I and this applies to all the chromium
based The process is very similar in other browsers with only minor
visual differences in the inspect element window. At first glance this might seem overwhelming but don't
worry just ignore the clutter and focus on the network tab. It looks like a wifi signal.
If the window fails crowded click the three dots then the icon that says unlock into
separate window to open it in its own space. Refresh your browser.
The network tab data will then populate. You can see this by clicking through the tabs.
Finding the correct relative URL requires some detective work. There is no single hard rule.
It's kinda an art honestly. But there are few tricks using them. You can streamline your investigation work.
So let's get into the tricks. Focus on the type column. Your target is to find a text document
right click and select inspect. A window will pop up. Hi guys. As you know for the past few weeks I have
that can be in any format like table format.
The page for this website I told a fresh use after a few minutes. If it isn't for your targeted website
don't worry just look for the rows in type column. If it says document
you have likely found it. Click on it. Then click on response tab.
This shows the data fetched by the website and it uses its own URL internally.
Examining the response details will confirm whether it is the correct data. In this case it looks like stock market data.
To find the magic URL itself click the headers tab. Hears tab is crucial for python request.
It provides data points to help the python data request as if it's coming from a browser.
The request URL is what you need here. Right click and copy it.
By the way for this video we don't need other header information. But at some point in your web scraping journey you might need it.
been working with Raspberry Pi. I have created a dedicated series on this and posted a few videos
Finding the relative URL was easy in this instance but that's not always the case. If your page doesn't auto refresh or if
you can't locate the relevant data especially on a large site it can get difficult.
These sites often contain a lot of data in various formats. JSON
Requiring significant scrolling to find the right document type. Instead of clicking through numerous tabs
the quickest and easiest method is usually to check the Fetch XHR tab first.
You will likely find the relative URL there. If not try the doc tab. The URL is almost
always in one of these two. You can safely ignore
Their names clearly indicate they wouldn't contain the data you need. The media tab might be helpful if you want to fetch video
or audio from the site. So that's less relevant to quantitative trading but maybe relevant to Quant traders you know.
What I mean? If the fetch and doc tabs don't yield results the WS short for WebSocket section is the
on how to get started. Now the initial part of the setup is complete and it is time to start writing our python
next place to look. This involves a slightly more complex process.
As the data is streamed in binary or string format depend on the website.
This section is useful for scraping tick data or interacting with the browser via WebSockets.
Actually many discount brokers' APIs use this method. brokers' APIs use this method. I might make a future video on fetching or interacting with
WebSocket in Python but that's a steeper learning curve for beginners. Master this one first and then we will
get into WebSockets later. Ok our data exploration often the trickiest and arguably the
most important part is complete. The more you practice the more familiar you become.
In my opinion creating your own web scraper is better than relying on Python libraries
for fetching the market data. I learned this the hard way your code built with such
libraries often breaks after a few months or years when websites update their user interface or underline structure or underline structure
scripts for algo trading. This video is operating system agnostic. No raspberry pi is needed.
or underline structure and is often happens far more. The rise of large language models has made data protection
incredibly difficult. Companies are realizing their data's value and are working to protect it from those who
would use it commercially. OpenAI this is understandable as it's unfair as I mentioned earlier in the disclaimer.
So if we hit the wall and the website blocks our request using Cloudflare usually. Luckily there are a few Python libraries
as a built in replacement for traditional 'requests' library to get past that firewall.

Open the web page you want to scrape in your browser. For this video I am going to scrape sgx

You know 8 or 9 years ago when I transitioned from discretionary to quantitative trading

response code in PyChimp's variable window here. A 200 response code means

the request was successful. indicates client error

indicates the request failed. Time to post process that FacedHTML data. Expanding the fetched data reveals a text

field containing an HTML formatted string. Beautiful soup then parses and cleans

this HTML using two lines of code. So pandas can process it with

df=pd.read_html(html_data) Pandas can convert this HTML

data into a neat data frame. Let's look at the special variable window again.

Here we can see that our data frame df actually has 9 different data frames within it.

If we expand it we can see that various data frames. Let's open the first data frame by clicking on View as data frame.

This is the same data frame that we can see in table format on the website. And if we click on the second data frame

I gradually realized that data is far more important than analytical skills.

it corresponds to the second table on the hgx nifty webpage.

If you want to fetch other indexes like the snp500 or laujones you can click on data frame

number 6 and get the relevant data. You can use the same code for various other symbols.

Since we need the hgx data we would use the first element in the df list.

Now if we open this hgx data frame a datetime index is missing. So we will create a column called dat.

Set it as the index and then export it as a CSV file. The remaining part of the code is optional.

Here I converted the hgx data frame into a human readable text format. Then used gtts and pygame

library to read aloud that text. Most examples for gtts use reading and writing files to the disk.

But here we wrote the files in virtual memory to speed up the process. And if you have noticed I am using a python environment

installed on raspberry pi via sssg in pycharm. If you want to know more about how to do that I have made a video on it.

I mean look at this way. There is an upper limit to how much analysis you can perform on a single data source.

Check that out. I will put the link in the description. If I execute these lines you will hear the tts engine.

I have connected the speaker to raspberry pi's auxiliary input. Sound quality might sound different

because my microphone is far away. Listen.

So we will conclude our video with this. Our job is done. We will put this script in a linux cron tab and if you are on

windows you can put the script in windows task scheduler. In the next video I will show you how I modified the investing.py

library using the same technique explained in this video. That library is down because of a cloudflare firewall.

We will use our adjusted script to grab the economic event data. After that we will feed the human readable version into

google's text to speech engine like we did here. So take care and happy coding. See you in the next one.

But imagine finding an underutilized data source and then applying similar advanced analytical methods and

seeking that crucial edge. Furthermore data has become crucial for machine learning or AI.

Therefore innovative ways to find data where others aren't looking are increasingly important.

We can't rely on traditional market data vendors for the same old generic data everyone else uses.

This is where web scraping becomes valuable. So let's get into a practical example. Our first web scraping

script will fetch SGX Nifty data. So this is the website we are going to scrape the data from.

The site uses other symbols so you can easily modify a small

nifty data as an example and I am using Edge a chromium based browser.

part of the code to use them instead. The basic structure of Python code for web scraping is

straightforward and largely consistent. The real trick lies in obtaining the

necessary URLs and headers so your script mimics like a web browser.

Please consider the following disclaimer before proceeding. One crucial thing to remember is the

legality of web scraping. It's generally acceptable to scrape

openly available data but don't redistribute commercially or collect personal information like

email IDs or user names. I mean there is a still grey area but avoid overloading the target servers.

This can lead to your IP being banned. Keeping these points

If you want to remember only one thing from this video then remember this. You simply have to right click on the

webpage you want to scrape the data from and select inspect element from the popup menu. That's all.

So the method is virtually identical to Google Chrome. Simply press Ctrl plus Shift plus I or

Alternatively use the keyboard shortcut Ctrl Shift I and this applies to all the chromium

based The process is very similar in other browsers with only minor

visual differences in the inspect element window. At first glance this might seem overwhelming but don't

worry just ignore the clutter and focus on the network tab. It looks like a wifi signal.

If the window fails crowded click the three dots then the icon that says unlock into

separate window to open it in its own space. Refresh your browser.

The network tab data will then populate. You can see this by clicking through the tabs.

Finding the correct relative URL requires some detective work. There is no single hard rule.

It's kinda an art honestly. But there are few tricks using them. You can streamline your investigation work.

So let's get into the tricks. Focus on the type column. Your target is to find a text document

right click and select inspect. A window will pop up. Hi guys. As you know for the past few weeks I have

that can be in any format like table format.

The page for this website I told a fresh use after a few minutes. If it isn't for your targeted website

don't worry just look for the rows in type column. If it says document

you have likely found it. Click on it. Then click on response tab.

This shows the data fetched by the website and it uses its own URL internally.

Examining the response details will confirm whether it is the correct data. In this case it looks like stock market data.

To find the magic URL itself click the headers tab. Hears tab is crucial for python request.

It provides data points to help the python data request as if it's coming from a browser.

The request URL is what you need here. Right click and copy it.

By the way for this video we don't need other header information. But at some point in your web scraping journey you might need it.

been working with Raspberry Pi. I have created a dedicated series on this and posted a few videos

Finding the relative URL was easy in this instance but that's not always the case. If your page doesn't auto refresh or if

you can't locate the relevant data especially on a large site it can get difficult.

These sites often contain a lot of data in various formats. JSON

Requiring significant scrolling to find the right document type. Instead of clicking through numerous tabs

the quickest and easiest method is usually to check the Fetch XHR tab first.

You will likely find the relative URL there. If not try the doc tab. The URL is almost

always in one of these two. You can safely ignore

Their names clearly indicate they wouldn't contain the data you need. The media tab might be helpful if you want to fetch video

or audio from the site. So that's less relevant to quantitative trading but maybe relevant to Quant traders you know.

What I mean? If the fetch and doc tabs don't yield results the WS short for WebSocket section is the

on how to get started. Now the initial part of the setup is complete and it is time to start writing our python

next place to look. This involves a slightly more complex process.

As the data is streamed in binary or string format depend on the website.

This section is useful for scraping tick data or interacting with the browser via WebSockets.

Actually many discount brokers' APIs use this method. brokers' APIs use this method. I might make a future video on fetching or interacting with

WebSocket in Python but that's a steeper learning curve for beginners. Master this one first and then we will

get into WebSockets later. Ok our data exploration often the trickiest and arguably the

most important part is complete. The more you practice the more familiar you become.

In my opinion creating your own web scraper is better than relying on Python libraries

for fetching the market data. I learned this the hard way your code built with such

libraries often breaks after a few months or years when websites update their user interface or underline structure or underline structure

scripts for algo trading. This video is operating system agnostic. No raspberry pi is needed.

or underline structure and is often happens far more. The rise of large language models has made data protection

incredibly difficult. Companies are realizing their data's value and are working to protect it from those who

would use it commercially. OpenAI this is understandable as it's unfair as I mentioned earlier in the disclaimer.

So if we hit the wall and the website blocks our request using Cloudflare usually. Luckily there are a few Python libraries

as a built in replacement for traditional 'requests' library to get past that firewall.